PROGRAM OF The 114thMeetingof the AcousticalSocietyof America Hyatt RegencyHotel © Miami, Florida ß 16-20 November1987 MONDAYEVENING,'I6 NOVEMBER1987 UM AUDITORIUM, Tutorial 7:00 TO 9:00 P.M. Lecture Introductionto Musical Acoustics.ThomasD. Rossirig(Departmentof Physics,Northern Illinois University,DeKalb, IL 60115) Musicalacoustics isa broadinterdisciplinary fieldthatdealswiththeproduction, transmission, andperception ofmusical sound. It overlaps several otherbranches ofacoustics, suchasphysical acoustics, shock and vibration, architectural acoustics, psychoacoustics, speech communication, andelectroacoustics. In thisintroductorytutorial,considerable attention willbepaidto howsoundisgenerated byvariousmusicalinstruments. In percussion instruments andin certainstringinstruments, theplayersupplies energy bystriking orplucking theprimaryoscillator (string,bar,membrane, orplate),whichin turntransmits energy tootherpartsofthe instrument.Wind instruments, on theotherhand,dependuponnonlinearfeedbackfrom theair columnto the inputcontrolvalveinordertogenerate sustained oscillations. Similarfeedback exists inabowed stringinstrument,andto a lesserextentin the singingvoice.Time will allowonlya briefdiscussion of howthe soundis transmitted fromthesource tothelistener viatheauditorium orbyelectroacoustic means, suchasrecording, reproducing, andbroadcasting systems. Finally,certainaspects of theperception of musicalsound(suchas loudness, pitch,andtimbre) will beconsidered. TUESDAY MORNING, 17 NOVEMBER 1987 BOUGAINVILLEA ROOM, 8:30 TO 11:05 A.M. SessionA. EngineeringAcousticsI: StructureInteractionand TransducerArrays AnthonyJ. Rudgcrs,Chairman UnderwaterSoundReferenceDetachment,Naval ResearchLaboratory,P.O. Box 8337, Orlando,Florida 32856 Chairman's Introductions8:30 ContributedPapers 8:35 AI. Application,ofa variationalprinciplefor fluid-structureinteraction to theanalysisof the response of an elasticdiskin a finitebaffle.Jerry H. Ginsbergand Pei-Tai Chen (Schoolof MechanicalEngineering, GeorgiaInstituteof Technology,Atlanta, GA 30332) pressure on the baffleradiusis examined;resultsin the limit of an infinite baffleareshownto agreewellwitha prioranalysis by AlperandIVlagrab [J. Acoust.Soc.Am. 48, 681-690 (1970) ]. [Work supportedby the Officeof Naval Research,Code 1132-F.] A variationalprinciplederivedfrom the Kirchhoff-Helmholtzinte- graltheorem waspreviously employed topredictthesurface pressure and vibrational response ofanunbaffled elasticdiskto harmonicexcitation[J. H. Ginsbergand A.D. Pierce,J. Aeoust.Soc.Am. Suppl.I 79, S35 (1986)]. Here,theanalysis isextended todescribe a finiterigidbaffleby usingassumed modefunctions for the pressure distribution alongthe vibratingandrigidsurfaces. Versionsusingvariousanalyticalfunctions, aswell asfiniteelementmodes,are developedand examinedfor convergence,numericalaccuracy,and efficiency.Dependenceof the surface $1 J. Acoust.$oc.Am.Suppl.1, Vol.82, Fall1987 8:50 A2. Modal contributions of a finiteplateto powerspectra.SungH. Ko (Naval Underwater SystemsCenter, New London, CT 06320) The purposeof thiswork is to comparethe modalcontribution of a finiteelasticplateto thepowerspectrumand the flexuralwavecontributionof aninfiniteelasticplateto thepowerspectrum.Flexuralwaw:sin an infiniteplatesubmerged in a fluidproducepressure wavesin thefluidthat 114thMeeting:Acoustical Societyof America $1 travelparallelto theplateanddecayawayfromtheplatesurface in the cylindrical NFCAshavebeendeveloped. Thispaperdescribes a spherical fluid.Thesewavesmaybecalledevanescent wavesandarestrongest in a regionneartheplatewheretheymayrepresent noise.Whena rectangular platesubmerged in a fluidisexcited,thevibratingplategenerates pressure wavesthat decayawayfrom theplatesurfacein thefluidandan acoustic wavethatradiatesintothefluid.The analyticalmodelconsidered hereisa two-sided fluid-loaded rectangular platesimplysupported at itsboundariesandexcitedby a forcingfunctionononesideof theplate.Theoretical analyses arediscussed andthewavenumber-dependent transferfunctions thatareusedforcalculating powerspectraarepresented. [Work supported by NUSC. ] NFCA fordetermining thefull three-dimensional radiationpatternof an enclosed transducer. In theory,theNFCA consists of a largenumberof discrete hydrophone elements arranged overa spherical surface inNequispaced constant latitudebands.In practice,thearraycanbesynthesized by useof a singlesemicircular arccontaining relativelyfewelements and 9:05 byeitherrotation ofthetransducer orrevolution ofthearcabout thepolar axis.Individualhydrophone sensitivities (bothamplitude andphase)are selected sothat theNFCA servesasa plane-wave filterfor acousticradiationfrominsidethesphere. Therelativesensitivities, calledshading coefficients,arecomputed usingtheNFCA reciprocityprincipleanda leastsquaresprocedure.Only a few setsof N/2 shadingcoefficients are requiredfor full three-dimensional operation of theNFCA overa frequencyrangeof threeoctav•sor more.Designcriteriaarediscussed and samplenumericalresultsare presented. A3. Impulseresponses andradiationimpedaneas of cylindricalvibrators usingwave-vector-time domain(k-t) methods. PeterR. Stepanishcn and D. D. Ebenezer(Departmentof Ocean Engineering,Universityof RhodeIsland,Kingston,R! 02881-0814) A wave-vector-timedomain (k-t) methodis presentedto evaluate the harmonicandtime-dependent loadingon a cylindricalshellthat is vibratingwitha specified spatialandtime-dependent velocity.The method is basedon utilizinga specified time-dependent modalexpansion for the radialvelocityof the vibrator.The acousticloadingon the vibratoris alsoexpressed asa modalexpansion in whicheachcoefficient isa summationofconvolution integrals of themodalvelocitycoefficients withmode- dependent radiationimpulse responses. In contrast toanearlierwork[ D. D. Ebenezerand Peter R. Stepanishen, J. Acoust.Soc.Am. 81, 854-860 (1987) ], the radiation impulseresponses are evaluatedusinga (k-t) methodbasedona time-dependent Green'sfunctionfor a baffledcylindrical vibrator.A comparison of the newmethodwith the earliermethodis presented alongwithnumerical results forvarious axialmodeshapes and circumferential modenumbers.[Work supported by ONR.] 9:20 A4. An array to produceconveerednormalvelocityandpressurefieldsfor wavenumbercalibration and boundarylayer modification.H. C. Schau, L. Dwight Luker, and S. Petrie (Naval Research Laboratory, Underwater Sound Reference Detachment, P.O. Box 8337, Orlando, FL 32856-8337) 9:50 A6. Designof an experimentto measurethe acousticmutualimpedances of a resonant,close-packedsonar array. Stephen C. Thompson (Acousticsand SensorSystems,OceanSystemsDivision,Gould, Inc., 18901 Euclid Avenue, Cleveland, OH 44117) Knowledgeof the acousticmutualimpedancecoefficients for a sonar array is necessary to achievean accurateanalyticaldescriptionof the performanceof the array. In the early stagesof a design,analyticalapproximations,e.g., of an array in an infiniterigid planebaffle,provide adequate estimates. Whenprototypehardwareisavailable,it maybeappropriateto deriverefinedestimatesof the impedancecoefficients. This allowsmoreaccuratepredictionof theeffectsof designchanges onperformance,and alsoprovidesan analyticalbaselinefor a designthat is enteringproduction.The measurement of themutualimpedance coefficients in a resonant,close-packed array is difficult.The only previouslyreported attempt[StephenC. Thompson,J. Acoust.Soc.Am. Suppl.1 68, S34 (1980] achievedinconclusiveresults,due to inaccurateestimatesof the elementelectromechanical transfermatricesand strongelementinteractionsthat occurin the vicinityof the elementresonance frequency.These measurements arcbeingrepeated,with greatercaregivento thepreviously noteddeficiencies. The experimentwill be described, concludingwith therequirements placedon thespecialtransducer elementsthat werenecessaryfor the measurement. The designof thesetransducerelementsv½ill be describedin a companionpaper in this session[M.P. Johnson,J. Acoust.Soc.Am. Suppl. 1 82, S2 (1987)]. A transmittingarray is describedthat is capableof producingspecifiednormalvelocitydistributions at thesurfaceof thearray.Suchan array is usefulin the productionof convected normalvelocitiesand pressure fieldsthat may emulateturbulentboundarylayer pressurefields.The pressure fieldsareconvective in nature,movingparalleltotheplaneofthe arraywitha phasevelocitythat isslow,i.e.,lessthanthespeedof soundin themedium.A slowphasevelocitycorresponds to a spatialwavenumber k thatisgreaterthanthecorresponding acoustic wavenumber inthemedium.Applications includeboundarylayermodification, testingof sensors in controlled TBL flows, and wavenumbercalibration of sensors.Results arepresented for severalprototypearraysinvolvingdifferentconstruction methodologies. Shadingschemes for theproductionof uniformconveered pressure fieldsare discussed relativeto the physicallimitationsof the deviceand electronics.Measuredpressurefieldsand phasevelocityare contrastedwith thosepredictedtheoretically.Plansfor producinglarger and morecomplexarraysare discussed relativeto currentprototyperesults.[Work partiallysupportedby ONT.] 9:35 AS. Sphericalnearfieldcalibrationarray for three-dimensionaiscanning. A. L. Van Buren (Underwater Sound Reference Detachment, Naval ResearchLaboratory,P. O. Box 8337,Orlando, FL 32856-8337) Nearfieldcalibrationarrays (NFCAs) are usedto determinethe farfield acousticradiation from a transducerby measurementsmade in the nearfield.The originalTrott NFCAs wereplanararrays.More recently, S2 d. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 10:0S A7. Design of a transducerelementfor an acousticmutual impedance measurementexperiment. Michael P. Johnson (Acoustics and Sensor Systems,Ocean SystemsDivision, Gould, Inc., 18901 Euclid Avenue, Cleveland, OH 44 117) A companionpaperin this sessionhas describedan experimentto measurethe acousticmutual impedancesin a resonant,close-packedso- nar array [S.C. Thompson,J. Acoust.Soc.Am. Suppl.182, S2 (1987) ]. This experimentrequiresthe useof specialmeasurement transducers in placeof theelements in thearrayunderstudy.Therewereseveralspecial requirements for thesetransducers: ( 1) highmechanical inputimpedance at the elementradiatingface; (2) accurateknowledgeof the transducer electromechanical impedancematrix;and (3) radiatingfacedimensions that are identicalto thoseof the clementsof the array understudy.The firstrequirementreducesthe effectsof interelementcouplingand,consequently,reducesthe sensitivityin the calculationsto errorsin the measurement.The secondacknowledges that the electromechanical impedance matrices of the transducer elements are needed in the calculation. The third requirementprovidesa measurementarray with the samemutual impedancecoefficients asthe array understudy.A prototypetransducerelementthat seemsto meettheserequirementsis a doublyresonant pistonelement.The elementisdesigned sothattheoperatingbandof the array understudyfallsbetweenthe two resonantfrequencies of the measurementtransducer. The designandperformance of thiselementwill be described. 114thMeeting:AcousticalSocietyof America $2 10:20 AS.A diffuse fieldsonartankformeasurement ofprojector sound power levelanddireetivity index.David Lubman(D. Lubman& Associates, 14301MiddletownLane,Westminster, CA 92683) Though widelyusedinairacoustics, thediffuse fieldtechnique hasnot beenextensively usedin underwater acoustics. Thispaperdescribes a small (2ma)diffuse fieldsonar tankfacility intended forproduction testingofthesteady-state sound-power levelanddirectivity indexof widebandsonarprojectors. Whenevaluated bystandards usedin air acoustics toensure adequate diffuse fieldmeasurement accuracy, thiscompact and inexpensive facilityappears tobequalified forthedetermination of bandaveraged sound.-power levelovera frequency rangeexceeding fiveoctaves thefactorsgoverning therealizations of practicalechosonde patterns; whicharesuitable for turbulence detection of atmospheric irregularities, suchasthermalstructures, dynamics,andturbulentvelocityfielctsat the loweratmosphere; andfromwhichphysically realizablesourcefieldsex- thesized. Last,a comparative analysis showsthattheapproximate pat- hibiting no abrupt discon acros the ante ape ca be sy ternsgenerated fromthe synthesized sourcedistributions manifestgood mean-square fitsto the idealizedacoustic beamsoriginallyspecified almostall overthe entirevisiblerangesof the patternsprescribed.[Work supported by LagosStateUniversity,BadagryExpressway, Ojo, Lagos, througha VisitingProfessorial Appointment. ] (fromabout4 lo 160kHz). Whencompared witha free-field measurementfacility, thepractical advantages ofthisapproach include largesaw ingsin costandspace. Longtransducer risetimesarepermitted since measurements aresteadystate.Moreover,theimplicitspatialintegration ofradiation patterns makes results ofdiffuse fieldtesting moreappropriatethanfree-field testingfor someapplications. 10:50 AI0. Development of higher-orderZernike polynomials employedin the analysis of aeonstie sensing antennas. S. AdeniyiAdekola(LagosState University(FETES), BadagryExpressway, Ojo,Lagos,Nigeria) Zernikecirclepolynomials havebeenfoundtobeinvaluableanalytical toolsforinvestigating thecharacteristics of acoustic sensing antennasIS. A. Adekola,Acustica47, 11½-131( 1981) ]. New analyticalresultsob- 10:35 A9. Practicalacousticbeamsandsynthesized sourcefieldsfor turbulence detectionat the lower atmosphere. S. Adeniyi Adekola (LagosState University(FETES), BadagryExpressway, Ojo, Lagos,Nigeria) tainedfromaninvestigation ofthegeneralized R • ( r ) Zernikecirclepolynomials, summarized in theabove-mentioned reference, areof higherdegreesof accuracythan thosepresentlyavailablein the literature.Indeed, tiesat theboresight regionoftheapertureandareconsiderably attenuated thesepolynomials aregenerated to 30øand29øof freedomdescribing the even-andodd-numbered modes,respectively; and the fieldradiatedby the acousticsensingantenna,whichis then expressed in termsof the polynomials, isevaluated throug.h directroutineintegration. Someof the newanalyticalresultsevaluatedand numericallycomputedare heredisplayedin graphicalformsfor completeness. Also,illustrative examples givenshownotonlyhowtheZernikepolynomials canbeemployed in the interpretations of theradiationcharacteristics of theacousticechosoundingsystems, but alsohowusecanbemadeof thepolynomials to describe towards the rim of the antenna cuff. It is also shown that if the beamwidth significantlytaperedsourcedistributionsessentialin successfulacoustic ofthepatternprescribed isextremelynarrow,thenthesourcedistribution synthesized fromit tendsto produce anundesirable strongfieldor singularity at therim of theantennacuff.The paperthenfocuses attentionon remotesensing of theloweratmosphere. [ Work supportedbyLagosState Computersimulations of newanalyticalresultsthat are important extensions of an earlierwork IS. A. Adekola,J. Acoust.Soc.Am. 76, 345368 (1984) ] are herepresented for the echosonde (acousticechosounding) system.It isshownthatthesourcedistributions synthesized fromthe prcassigned dinretirepressurefieldsarc not only confinedwithin finite antennaapertureregions,butarealsocharacterized by maximumintensi- TUESDAY' MORNING, 17 NOVEMBER 1987 UNIVERSITY University, BadagryExpressway, Ojo,Lagos,througha VisitingProfessorialAppointment.] LECTURE HALL, 8:30 A.M. TO 12:00 NOON Session B. Psychological andPhysiological Acoustics I: Hearing-Impaired Speech andAuditoryPerception JosephW. Hall, III, Chairman Division of Otology, University ofNorthCarolina, 610Burnett-Womack Building, 22911,Chapel Hill, North Carolina 27514 ContributedPapers 8:30 B1. The effect of varying the amplitude-frequency responseon the maskedSRT for sentences in hearing-impairedlisteners.JanctteN. van Dijkhuizen, Joost M. Festcn, and Reinlet Plomp (Experimental Audiology,ENT Department,Free UniversityHospital,Amsterdam, withinlimits,unaffected bydynamic variations in theamplitude-fi'equencyresponse ofthehearingaid.Fornormal-hearing listeners, vanDijkhui- incomingsignal,. mayoptimallydeliverspeech to an impairedcar.Sucha zen et al. [J. Acoust.Soc. Am. 81, 465-496 (1987)] found that the maskedSRT for sentences is remarkablyresistantto dynamicvariations in the slopeof the amplitude-frequency response whenit shapedboth speechand noise.In thisexperiment,we studiedcorresponding conditionsfor hearing-impaired listeners. Again,the amplitude-frequency response shapedbothspeech andnoiseandthusleft speech-to-noise ratios untouched. The resultsobtainedso far alsoindicatethat, for hearingimpairedlisteners, dynamicvariationsof the amplitude-frequency re- systemrequiresthat the speech-reception threshold(SRT) in noiseis, sponsedo not affectthe SRT. The Netherlands) A multichannelgain-controlhearingaid, in which the frequency-dependentamplificationis adaptedautomaticallyto the fluctuationsof the S3 J. Acoust. Soc.Am.Suppl.1,Vol.82, Fall1987 114thMeeting: Acoustical Societyof America S3 8:45 B2. Explorations on the differencein $RT betweena stationarynoise masker and an interfering speaker.Joost M. Festen (Experimental Audiology, ENT Department, FreeUniversity Hospital,Amsterdam, The Netherlands) For moderatehearingimpairment,the speech-reception threshold (SRT) for sentences in stationarynoiseisroughlyup to 5 dB higherthan correlations werefoundamongthepsychoacoustic variables, whichalso predicted PFFIN score. Correlations between hearing levels andPFFIN remained significant afterpartiallugage,noiseexposure, upwardspread ofmasking at2 kHz,andeducational andlinguistic measures. Systematicallytruncatingthe HL distribution showedthat thiscorrelation lies mostlyin the 15-25dBHL rangeandimplicates low-aswellashighfrequency hearing. Thusthe17%oftheadultpopulation between 15and 25 dB HL (0.5-4.0 kHz) shouldbe consideredmarginal,both clinically fornormalhearing.However,a fluctuating interfacing soundseems more typicalfor everydaylisteningconditions thanstationary noise.For a groupof younghearing-impaired listeners, thedifference in SRT with normal-hearing listenersincreased from 2.7 dB for stationarynoiseto 7 dB for 100%sinusoidally intensity-modulated noise,andevento 10.5dB forinterfering speech. In thisexperiment, theinterfering speech wasthe time-reversed voiceof the samespeakeras the signal,whichmay have introducedan extradifficulty.In a subsequent experiment,the SRT for sentences fromtwo speakers (onemaleandonefemale)wasmeasured witha varietyof fluctuating maskers covering therangefromstationary noisetointerfering speech. Thetwovoices weremutuallyusedasmusket, givinga meaningful interference. Meaningless interference wasobtained bytimereversal oftheinterfering speech orbymodulating noisewiththe envelope of speech. Two typesof modulation wereused:onebroadband andonein whichhigh-and low-frequency noise(separationfrequency 1000Hz) wasmodulatedwith the high-andlow-frequency envelope of speech,respectively. 9:00 B3. Reducedauditory discriminationand compression in hearingaids. Wouter A. Dreschler(AcademicMedical Center,Clinical Audiology, University of Amsterdam, Amsterdam,The Netherlands) andexperimentally. Scientifically, if notpragmatically, a "lowfence"at 15dB HL isjustified. 9:30 B$. EvaluatingtheArticulationIndexfor auditory-visualinput.Ken W. Grant (ResearchLaboratoryof Electronics, Massachusetts Instituteof Technology, Cambridge, MA 02139) The currentstandardfor calculatingthe ArticulationIndex (AI) ineludesa procedureto estimatethe effectiveAI whenhearingis combined withspeechreading [ANSI S3.5-1969(R 1978), "Methodsfor theCalculation of the Articulation Index" (American National StandardsInsti- tute, New York, 1969)]. This procedureassumes that the band-importancefunctionderivedfor auditorylisteningsituationsappliesequally well to auditory-visualsituations.Recentstudieshaveshown,however, that certain auditory signalsthat, by themselves,producenegligible speechreceptionscores(e.g.,FO, speech-modulated noise,etc.) canprovidesubstantial benefitsto speechreading. The existence of suchsignals suggests that the band-importance functionfor auditoryand auditoryvisualinputsmaybedifferent.In thepresentstudy,an attemptwasmade to validate the auditory-visualcorrectionprocedureoutlined in the ANSI-1969 standardby evaluatingauditory,visual,andauditory-visual sentence identification performance of normal-hearing subjects for both Two specificapplicationsof single-channel compression are studied separately: whole-range compression asa compensation for recruitment, andcompression limitingasan elegantmethodto limit excessive output wideband speech degraded byadditivenoiseandbandpass-filtered speech presented in quiet.The resultsobtainedfor auditorylisteningconditions withanAI greaterthan0.03supporttheprocedure outlinedin thecurrent levels.In casesof compression limiting, compression provedto besuperior to peakclipping,especiallyfor the perceptionof consonants: An averageimprovementin identificationscoresof 18% wasfound and lessrolloverfor the perceptionof finalconsonants at highpresentation levels.In casesof whole-rangecompression, a small, but significant,increasein identificationscoreswasfoundfor settingswith the compression kneepointat the lowestlevel.A detailedanalysisof the patternsof confusions revealedclearqualitativedifferences in the perceptionof phonemes with and withoutcompression. Thesedifferences canbebroughtinto relation ANSI standard.[Work supportedby NIH.] with an improvedperception of temporalCues,like the prebursts/lent 9:45 B6. Frequencydiscriminationand consonant recognitionin the hearing impaired.Marleen T. Ochs (Departmentof CommunicationSciences and Disorders,RadfordUniversity,Radford,VA 24142) The frequency difference !imen(DL) for short-duration second-formant transitions(F2) presentedaloneand in speechlike environments was examinedin normal hearingand hearing-impaired subjects.Four intervalof plosives,and with spectralchangesdueto the activationof the compressioncircuit. stimulusconditionswere included:F2 transitionalone;F2 transition in thepresence of all formants(full formant); full-formantstimulipreceded by a hurstand followedby a vowel;andfull-formantstimuliprecededand followedby a vowel.For the simpleststimuluscondition(F2 alone), all 9:15 hearing-impaired subjects hadDLs withinthe 95% confidence interval B4. Correlationsamongauditory functionsin the "normal" range. around the mean of the normals. For all other conditions,however,the Mark Haggard,GabrielleSaunders,and StuartGatehouse(MRC Institute of Hearing Research, Nottingham NG7 2RD, United hearingimpaireddemonstrated extremeintersubject variation.Some hearing-impaired subjects couldnotdetecta formanttransition of even Kingdom) 800Hz, whereasotherscontinuedto performlike normalhearers.Consonantidentification performance wasalsoexaminedin CV andVCV envi- To date,hearinglevelsandpyschoacoustic measures haveshownfactor structureandcorrelations with speechperceptionin noiseamongthe ronments. Consonants whose identification is believed to involve formant hearingimpaired, butnotamongclinically"normal"ears.A pseudofree- transitionswere selectedfor study.In spiteof modifications to the fre- fieldsentence-in-noise test(PFFIN) from a dummy-headrecording,sim- quencydiscrimination taskanduseof relevantsyllables, relativelyweak plifiedpsychoacoustic tuningcurves,andtestsof linguistic processing correlations continue to be observed between F2 DL and consonant rec- wereadministered to 60 peoplewith hearingbetterthan20 dB HL (0.52.0 kHz); one-thirdhad complainedof auditorydifficulties.Significant ognitionability, as well as betweenF2 DL and degreeof hearingloss. [Work supportedby NIH. ] 10:00-10:15 Break S4 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 114thMeeting: Acoustical Society ofAmerica S4 10:15 11:00 B7. Lateral suppression and auditory frequencyselectivity.Wouter A. Dreschlerand A. gens Leeuw (AcademicMedical Center,Clinical Audiology,Universityof Amsterdam,Amsterdam,The Netherlands) B10. Sinusoidal modeling of speech: Its use in simulating and For measuring frequencyselectivity,thenotched-noise procedure, introducedby Pattersonand Nimmo-Smith,is regardedat the momentas thebestmethodto avoidcueingeffects andto takeoff-frequency listening intoaccount. In thisstudy,a similarapproach wasadoptedfor themeasurement oflateralsuppression. The amountof suppression for a two-tone masker,asa functionof the frequencyof the suppressor, wasmeasuredin the presenceof a notchednoise.Also in this paradigm,a fair amountof suppression wasfoundin a groupof 15normal-hearing subjects. For hearing-impairedsubjects, and,surprisingly, alsofor subjects with a conductive loss,a graduallossof suppression as a functionof hearinglosswas observed. Finally,theresultsarerelatedto auditoryfiltershapes obtained in simultaneous andforwardmasking. compensating for recruitment of loudness. Janet R. Lias and Mark A. Clements(Schoolof ElectricalEngineering,GeorgiaInstituteof Technology,Atlanta,GA 30332) Amplitudecompression techniquesutilizing both singleand multichannelfilteringsystems havebeenproposed for usein compensating for recruitmentof loudnessin hearingaids.Due to a varietyof limitations, thesetechniques haveshownonlymarginalsuccess. The sinusoidal model of speechdevelopedby Quatieri and McAulay [IEEE Trans. Acoust. SpeechSignalProcess. ASSP-34,744-754 (1986) ] hasbeenusedto simulateandcompensate for recruitment,usingthegeneraltheorydescribed by Villehut [J. Acoust. Soc. Am. $6, 1601-1611 (1974)] and others. Specifically,the modelsynthesizes speechas the sum of sinusoldswith time-varying amplitudes andphases. Amplitudecompression andamplificationcanbeperformed onindividualsinusolds to shapespeech 'tofit an impairedperson'sresidualhearing.Conversely, expansion canbeusedto Simulate the effects of recruitment. This model offers almost unlimited 10:30 flexibilityin areassuchasconsonantboostingand spectralshapingin any portionof the spectrum.With a real-timeimplementation of thismodel, the parameters canbeeasilyadjustedto suiteachperson'sneeds. BS. Frequency masking patterns and "excess masking" in normalhearing and hearing-impairedlisteners. Margery A. Garrison, Anna C. Schroder,and David A. Nelson (Hearing ResearchLaboratory, Universityof Minnesota,2630UniversityAvenueS.E.,Minneapolis,MN 55414) 11:15 The presence of "excessive upwardspreadof masking"in high-frequencysensorineural hearing-loss(HFSNHL) subjectshas beenpreviouslydemonstrated byotherinvestigators. Thisexcess masking,defined by predictionsfrom a "noise"modelof hearingloss,occursin the transitionregionbetweennormalandimpairedhearing.To examinethisphenomenonfurther, frequencymaskingpatterns(FMPs) were obtained from two normal-hearingsubjectsand one HFSNHL subject.Using a 200-Hz narrow..band noise(NBN) masker(90 dB SPL) with an upper edgeat 520Hz, FMPs fromthe impairedeardemonstrated excessmasking,whichit washypothesized mightbeexplained by an inabilityto "listenin the valleys"of the maskerenvelopebecause of poortemporalresolution. To test this notion, the experimentwas repeatedusinga 40-Hz NBN masker(90 dB SPL), which has lessrapid envelopefluctuations. No differences in the upwardspreadof maskingwereseenbetweenthe 200-and40-Hz bandwidth-equal SPLmaskers.Whentheexperimentwas repeated withequalspectrum levelmaskers, the40-Hz NBN maskerproducedlessupwardspreadof maskingin bothnormal-hearingand hearing-impairedears.Thesedata suggestthat, for the HFSNHL subject, excessive upwardspreadof maskingcannotbecompletelyexplainedbyan inabilityto "listenin thevalleys."[Work supported by NINCDS.] B11.Asimpleprocedurefor measuringiso-loudnesscontuurs.$. B. Allen (AT&T Bell Laboratories,Murray Hill, NJ 07974), P.S. Chien (Center for Researchin Speechand HearingSciences, City Universityof New York ), andJ. L. Hall (AT&T BellLaboratories,Murray Hill, NJ 07974) A procedurefor measuring iso-loudness contoursin normalandhearing-impairedsubjects similarto thatusedby GellerandMargolishasbeen devised[J. SpeechHear. Res.27, 20-27 (1984) ]. The stimuliareburstsof noise«-octwide,centered at 1-octintervals from250Hz toto4 kHz. The subjectisaskedto ratethestimuliaccordingto a six-pointscale,usingthe descriptivewords "very soft," "soft," "OK," "loud," "very loud," and "too loud." The instructionsincludea one-sentence descriptionof each loudness category.Failureto respondis recordedas "nothing."In an initial pass,the lowerand upperlimitsof the subject's hearingrangeat eachfrequencyare estimated.Theseinitial trials alsoserveto train the subject.Following the initial pass,15 levels,logarithmicallyspacedbetweenthe measuredlimits at eachfrequency,are presentedin random order,threetimeseach.The entireproceduretakesabout30 min to com- plete,and it providesstableestimates(basedon test-retestmeasurements)of loudnessgrowthfunctions.Theseloudnessgrowthfuactions are then replottedas iso-loudness contours.Data from normal and hearing-impairedsubjectswill bepresentedalongwith their pure-tonethresholds. 10:45 11:30 B9. High-frequency audiomerry in platinum ototoxicity. Wouter A. Dreschler and Rob J. A.M. v/d Hulst (Academic Medical Center, Clinical Audiology, University of Amsterdam, Amsterdam, The Netherlands) In a groupof 144subjects(288 ears)medicallytreatedwith platinum drugs,high-frequency audiometrywas appliedat frequentintervalsin orderto comparetheeffectsof threedifferentdrugadministrations: highdoseels-platinumcures,low-doseels-platinumcures,and carboplatin cures.In all subgroups,high-frequencyaudiometryconsiderablyenhancedthe early detectionof ototoxicity.Marked differencesbetween cureswereestablished, bothin the patternof onsetof the damageand in the relationbetweendoseand damageseverity.For subjectstreatedwith platinumderivatives,especially,the thresholdsat 14 kHz prove to be important.The resultssuggest that, for thesesubjects, a singlemeasuring frequencyshouldbetakenintoconsideration. With a minimumof effort, mostof theincreased sensitivity fora complete high-frequency audiogram canbe reached.Finally,the predictivevalueof a thresholddeterioration above8 kHz for a thresholddeteriorationin the conventionalrangeof audiometricfrequencies will be considered. S5 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 B12. Effectsof fluctuatingconductivehearinglossin infancyon auditory brain-stem responsesand binaural interaction measured at age 5. Adele Gunnarsonand Terese Finitzo (University of Texas at Dallas/ Cailler Center, 1966 Inwood Road, Dallas, TX 75235) The purposeof thisresearchwasto examinelong-termauditoryelectrophysiologic effectsof earlyconductivehearingloss.Animal deprivation studiessuggestbrain-stemanatomicand physiologicalterationduring criticalperiods.Comparablehumanresearchis limited.The research questionedwhetherauditorybrain-stemresponses (ABR) and binaural interaction(BI) at ages5-7 differeddependingon documentedpatterns of earlyhearing.Stimuliwererarefactionclicksat 21.1/s, presented binaurallyat 60 dB and monaurallyat 30, 60, and 80 dB HL. The BI was derivedby subtractingsummedipsilateraland contralateralmonaural responses frombinauralresponses, anddefinedasdeflections whoseamplitudeexceeded 3 s.d.fromderivedprestimulus amplitudes. ResU{ts indicatedthat wavesIII and V, and I-lII and I-V interpeaklatencies(IPL) differedwhencontrolswerecomparedto experimentalsubjects. WaveI, the latency/intensityfunctionsat 30-60 and 30-80 dB, and III-V IPLs 114thMeeting:Acoustical Societyof America S5 The contribution of two acousticcues,voice onsettime (VOT) and wereunchanged. TheBI waspresent in 8 outof9 controls and8 outof9 childrenwithfluctuating mildlosses in infancy.Only4 outof 9 children vowelonsetfrequencies, to the perceptionof a TAD-DAD voicingcon- withmoderate asymmetric losses in infancy hadBI at age5. [WorksupportedbyNIH No. 5 ROI NS 19675-03.] trustwasexamined fortennormal-heating andsixhearing-impaired children (3 FA: 50-95 dB HL) between6 and 12 yearsof age.Bothnatural stimuliandhigh-qualitysynthesized copiesof thesestimuliwereusedto createfiveVOT continua(three naturaland two synthetic),in whichthe vowelonsetcuewaseitherappropriate or conflicting. An adaptiveidenti- 11:45 ficationprocedurewasusedto determinephonemeboundaries under B13. Cue saliencein the perceptionof a stopvoicingcontrastby hearing and hearing-impaired children. Valerie Hazan (Department of Phonetics and Linguistics, UniversityCollegeLondon,4, Stephenson Way, London NWl 2HE, United Kingdom), Lisa Holden-Pitt, Sally Revoilc,Donna Edward,andJanetDrogc(Centerfor Auditory andSpeechSciences, GallaudetUniversity,Washington, DC 20002) TUESDAY MORNING, 17 NOVEMBER thesefive test conditions.Changesin spectralcharacteristics at vowel onsetdidproducesignificant shiftsin phoneme boundaryfor thenormalhearinggroup,but notfor thehearing-impaired group.Similarphoneme boundaries wereobtainedby the normal-hearing groupfor naturalcon- tinuaandtheirsynthetic counterparts. However,highervaluesof standard error were observedfor syntheticstimulifor both listenergroups. UMS 4 AND 5, 8:30 TO 11:55 A.M. 1987 Session C. Underwater Acoustics I: Arctic Acoustics I Dan 1. Ramsdale, Chairman Naval OceanResearch and Development Center,NSTL Station,Mississippi 39529 Chairman's Introductlon•:30 Invited Papers 8:35 C1. Geoacoustics of the shallow-water Arctic. $. E. Matthews(NumericalModelingDivision,Naval Ocean Research andDevelopment Activity,NSTL, MS 39529-5004) Fromtheperspective of geology, theshallow-water regionsof the ArcticOceancanbedividedintotwo provinces of approximately equalsize.Oneis characterized byungiaciated, thick,prograding sedimentary deposits, whicharcfoundonthecontinental shelves of theBeaufort,Chukchi,andEastSiberian Seas.The otherischaracterized by a glaciated bank-trough terrain,andincludes thecontinental shelfof theCanadianGreenland ArcticandtheBarents,Kara,andLaptevSeas.Froma low-frequency geoacoustics pointof view, theprograding shelfprovince consists ofa fewtotensofmeters ofpoorlysortedelastic sediment ( 1.5-1.6kin/ s) that overliesan acousticbasementof consolidated coarsesediment( 1.7km/s). The bank-troughprovince hastwo distinctlydifferentenvironments: shallowbandsthat consistof zero to severalmetersof'coarse sediment ( 1.6-1.7kin/s) overlying anacoustic basement of sedimentary rock(3.5kin/s), anddeeptroughs flooredby a fewtensof metersof poorlysortedsediment (1.5-1.6kin/s) overan acoustic basement of consolidated sediment( 1.9km/s). Thesegeneralizations mustbeconsidered tentativebecause of thecritical scarcity ofdata;however, plane-wave reflection andnormal-mode calculations arepresented forthesegeneric environments to demonstrate their acoustic similarities and differences. 9:00 C2. Sound-speed structureof the Arctic Oceanandthe GreenlandSeaMarginal Ice Zone.JohnL. Newton (Polar ResearchLaboratory,Carpinteria,CA 93013) Within the centralice-coveredArctic Basin,well awayfrom marginalinfluences,the verticalsound-speed profilecanbecharacterized by fourpositivesound-speed gradientlayers.Theselayers(andtheapproximate gradients ) area sutracemixedlayerextending toa depthofa fewtensofmeters( • 0.016/s); a layerfromthe baseofthemixedlayertothedepth(200-500m) ofthetemperature maximumin theAtlanticwater( • 0.080/ s); a layerfromthetemperature maximumto thebottomof theArianticwaterat about1000m ( •0.010/s); andthedeepwaterextending to theoceanbottom( • 0.016/s).Theconsistent andnotabledifferences evident in thegeneralsound-speed structure of theCanadian andEurasianBasins canberelatedto thelargescale ArcticOceancirculation.Profile-to-profile variabilityandtheoccurrence of smallverticalscalesound-speed features,suchas relativeminima/maximaand steplikestructures, are significantly higherin the Eurasian Basinthanin theCanadianBasin.Thesound-speed distribution withintheGreenlandSeaMarginalIceZoneis dominated bythepresence of a strongoceanographic front,theEastGreenland PolarFront,whichmarksthe boundarybetweenthecool,freshPolarwaterexitingtheArcticOceanandtheadjoiningwarmer,moresaline Atlanticwater.Because ofthisstrongoceanfront,profile-to-profile sound-speed variabilityin theMarginalIce $6 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S6 Zoneishighovershortdistances andtimes.Thecomplex sound-speed structure anditshighvariability result fromvariousinteractions between thetwowatermasses whichcantaketheformofeddies, intrusive layers,and steplikefeatures. 9:25 C3. Recentadvancesin Arctic high-frequency acoustics.R. œ. Francois(Applied PhysicsLaboratory, Universityof Washington,Seattle,WA 98195) In thepastfewyears,research in Arctichigh-frequency acoustics (f> 5 kHz) hasbroughtabouta much betterunderstanding of theeffectof environmental factorson acoustics. Our knowledge of theeffectof water properties continues to increase aswell-knownphenomena areroutinelytreatedin greaterdetail.Ice properties,on theotherhand,andtheeffectof thesepropertieson acoustics, requirenewideasanddiligentresearch. For example,and perhapsof foremostimportance,considerreflections from seaice.Studiesof the reflecting propertiesof the lowersurfaceof seaice,with its dendriticlayer,showthat at low frequencies the reflection coefficient at normalincidence depends onlyon thebulkproperties of themedium.At higherfrequencies (up to 200 kHz) Gaussiandistributionsfor reflection coefficientsare observed,with standard deviationsthat increase with frequencyandaverages that decrease with frequency,asmuchas28 dB belowthat predictedby bulkproperties alone.Studiesof thetargetstrengthcharacteristics of iceblockshapes haveshownthat rigid body theoryfor simpleshapes,while goodfor hard ice surfaces,only crudelypredictsactualresponsefor freezingsurfaces. As for in situicekeels,thevariationin returnwith frequencyshowsthat reflections areoften the coherentsumof severalreflectorsnearthe samerangeand closeenoughin azimuththat suchreturns appearto bepointlike.[ Work is supported by Officeof NavalTechnology. ] 9:50 CA.A reviewof appliedArctic acousticmodels.R. C. Cavanagh,J. A. Davis (PlanningSystems,Inc., 7925 WestparkDrive, McLean,VA 22102), andJ. S. Hanna (SAIC, 1710GoodridgeDrive, McLean,VA 22102) Basicacousticmodeldevelopmentrelatedto the Arctic environmenthas beencarriedout by various researchers over the past25 years.This work is beingincorporatedinto appliedmodelsfor propagationand noisethat attemptto provideusefulpredictivecapabilitygivensuitableenvironmental databases.A reviewis presented ofthecandidatetheoriesandsomehistoryof theirdevelopment. In particular,icescattering theories aredescribed thatshowpromise ofutilityandtheirincorporation intotheseveralstandardpropagation models isdiscussed. Somecomparisons of thesemodelswith dataare presented and the outstanding issuesrelatedto modeldevelopment and the adequacy of existingdataarepresented. Althoughthe development of a correspondingambient noisemodel is a much lessmature activity, current attemptsat such developmentare reviewed.[Work supportedby the ASW EnvironmentalAcousticSupportProgram.] 10:15 C5. High-frequency acoustic backscatter from the Arctic ice canopy: Theory and experiment. Suzanne T. McDaniel (Applied ResearchLaboratory,PennsylvaniaStateUniversity,P.O. Box 30, StateCollege,PA 16804) The methodof physicalopticsisappliedto treatthebackscatter of high-frequency acousticsignalsfromthe Arctic ice canopy.Theoreticalresultsare obtainedfor both the meanbackscattering strengthand the spatial coherence of the scatteredfield.The meanbackscattering strengthis foundto bedependenton the fractional surfacearea coveredby underwaterice ridges.Using measuredice distributions,the variabilityin mean scattering strengthis predictedto varyby at most + 2.5 dB from its averagevaluefor a widevarietyof ice conditions. The theoryis thenappliedto treatthe spatialcoherence of backscatter. The coherence is foundto dependon the ice draft distributionand ensonified surfacearea.The resultsof experimentsto measurethe verticalspatialcoherence of bistaticbackscatter arepresented. Theseresultsnotonlyconfirmthetheoretically predicteddependences on measurement geometrybutalsosupportpredictions of therelativeimportanceof ice roughness versusensonification. [Researchsupported by ONT with technicalmanagement by NORDA.] 10:40 C6. Energy partitioning for very-low-frequency,long-range acoustic propagation in the Arctic. A. B. Baggeroer(Massachusetts Instituteof Technology,Cambridge,MA 02139), G. L. Duckworth (Bolt Beranekand NewmanLaboratories, Cambridge, MA 02238),andE. K. Scheer(WoodsHoleOceanographic Institute, Woods Hole, MA 02543) In long-rangeacousticpropagationin the Arctic, very-low-frequencysignalsinteract with both the ice canopyandthe bottom,bothof whichintroduceloss.Signalsfrom a sequence of long-range(340-kin) shots detonatedat severaldepthswererecordedon a multichannelarray and resolvedaccordingto their horizontal phasevelocityand frequencycomponents to separatethe losseffectsof the thesetwo boundaries.Sourcelevels andgeometric losses basedupona determined sound-speed modelweresubtracted to arriveat an absorptive/ scatteredlossmeasurefor eachresolvedpath and frequency.Thesereduceddata were usedto estimatethe S7 J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 114th Meeting:AcousticalSocietyof America S7 partitioningof energyaccordingto phasevelocityand frequencyfrom the icecanopyand the seafloor.At frequencies below20 Hz, bottominteractingsignalsaccountfor morethan 50% of the receivedenergyobservedat the array on almostall the shots. 11:05 C7. Arctic ambientnoisemechanisms. Ira Dyer (Massachusetts Instituteof Technology,Departmentof OceanEngineering, Cambridge, MA 02139) The studyof ambientnoisecan bedividedintotwo categories, onedealingwith thecorrelationof averaged noisewith the slowevolutionof environmental parameters,and the otherconcentrating on detailsof the ice fractureprocesses that ultimatelygiveriseto individualnoiseevents.Recentprogressin both categories, especially atlowfrequencies (3-100Hz) where significant results appear tobeathand, isreviewed. Thoughts onhigherfrequencies ( 100-3000Hz) arealsogiven,forwhichit seems likelythattheenvironmental correlates anddetailedsourcemechanisms areof a differentcharacterthanat low frequencies. Finally,distinctions are drawnbetweenpackiceof thecentralArcticandlow-concentration iceof themarginalicezone,withrespectto their likely mechanisms of noisegeneration.[Work supportedby ONR, Arctic ProgramOffice.] 11:30 C8. Arcticdatabuoys--Anul•late.Beaumont M. Buck(PolarResearch Laboratory, Inc.,6309Carpinteria Avenue,Carpinteria,CA 93013) A 1983paper,"Arcticdatacollection theeasyway--Data buoys"[J. Acoust.Soc.Am. Suppl.1 74, S2 (1983)], presented data buoysas a partialsolutionto the problemsand limitationsinherentto manned operations in theArctic.Thepresent paperdescribes thesuccesses andfailures ofthesedevices, asappliedto Arcticunderwater acoustics, in theintervening yearsandwhathasbeendonewiththelargequantities ofdata that theyhavegenerated. Thiswill includewide-area,synopticcollectionof ambientnoiseleveldata,aswellas first-ever measurement of low-frequency propagation duringthesummertime in the CentralArcticusing autonomous cwprojectorandreceiving databuoys.Plansforevenmoresophisticated devices to beinstalledin 1988arediscussed. [Work supported by U.S. Navy.] TUESDAY MORNING, 17 NOVEMBER POINCiANA 1987 ROOM, 9:00 TO 10:40A.M. SessionD. Architectural AcousticsI: Acoustical Design of Renovated Performance Facilities I ChristopherJaffe,Chairman Jaffe Acoustics, Inc., 114A WashingtonStreet,Norwalk, Connecticut06854 Chairman's Introductions9:00 Inoited Papers 9:10 D1. The movie!•alaceas!•erformingartscenter. PeterJ. George(PeterGeorgeAssociates, Inc., 34West 17th Street, New York, NY 10011) Many movie theaters in downtown neighborhoodshave been saved from demolition by their renovation andrecyclingasperformingartscenters.While thesespacesoftenhavephysicalcharacteristics that aredeemed acoustically desirable,thereare oftenmanyfeaturesthat are detrimentalto a goodacousticalperforming environment. Thispaperdescribes typicalattributesanddeficiencies withmethodology to improvethelatter. 9:40 D2. SomeunusualrenovationI•rojects.David L. Klepper (Klepper Marshall King Associates, Ltd., 7 Holland Avenue, White Plains, NY 10603) A comparison of several differentrenovation projects indicates thatanacoustical consultant mustbevery flexible in adopting a particular set of recommendationsfor the acoustical,architectural, and restoration S8 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 114thMeeting:AcousticalSocietyof America S8 requirements of a particularproject.In oneproject,sound-reflecting panelsmay be appropriate,in another, definitely not.In oneproject,an idealcentralclusterloudspeaker system maybeappropirate, in a second, it mustbesupplemented by delayedunderbalcony loudspeakers, whilea third wouldbestemployverticallinesourceloudspeakers, a fourtha horizontallinesource,anda fiftha distributeddelayedloudspeaker system.By meansof slides,the differences betweentheseseveralprojects,wh;.chincludea universityconcerthall, the conversion of a motionpicturetheatreto a multi-useperformingarts centerwith emphasison symphony concerts,a universitytheatre,and two 19thcenturyoperahousesrestoredasmulti-useperformingartscenters will beexplored. Onecommonthreadrunsthroughall of them,however; all makemaximumuseof livesound beforethe useof electronics,with minimumuseof sound-absorbing treatment. 10:10 D3. Adaptivereuseof existingauditoria.J. ChristopherJaffe(JaffeAcoustics,Inc., 114A WashingtonStreet, Norwalk, CT 06854) The adaptivereuseof existingauditofiais an extremelycostefficientway for communities to create contemporary "stateof theart" performance facilities. Thekeytothesuccessful culmination of sucha facility isa cleardefinitionof theprogramuseofthespaceandanunderstanding between theclientandtheconsultant thatall availableacoustic techniques canbeutilizedto achievefinaldesigngoals.Thispaperwill discuss the adaptivereuseof a numberof buildings asdiverse asa library,a unionhall,a WPA builtart decomusichall, anda threetheatre/filmcomplex. TUESDAY' MORNING, 17 NOVEMBER TUTTLE 1987 SOUTH ROOM, 9:00 TO 11:20 .A.M. SessionE. Noise I: Propertiesand New Applicationsof PorousMaterial Gilles A. Daigle, Chairman Divisionof Physics, NationalResearchCouncil,Ottawa,OntarioK1A OR6,Canada Chairman's Introduction--9:00 Invited Papers 9:05 El. Theoriesfor soundpropagationin rigid-framedprous materials. Keith Attenborough(Engineering MechanicsDiscipline, Faculty of Technology,The Open University, Milton Keynes MK7 6AA, United Kingdom) An historicaloverviewof thedevelopment of theoriesfor soundpropagationin rigid-framedporousmedia is offered.Phenomenological and microstructuraltheoriesare contrasted.Microstructuralapproaches may starteitherfromspecifications of thegeometrical formof thesolidframeor fromspecifications of thegeometri- calformofthepores. Theformerapproach isparticularly useful fordescribing acoustical properties offibrous materials.The latter approachhasthe longesthistory,the widerapplication,and is the basisof manyrecent developments in thedynamictheoryof poroelasticity. In oneformof thepore-based theory,consideration is givento the incorporation of dynamiceffectsof poreshapeand irregularityandsteadyflow tortuosity.An alternativeform resultsfrom a morerigorousapproachto the effectson viscousdragand thermalinteraction with porewallsof the inclinationof poreaxesto the macroscopic pressuregradient,i.e,, dynamictortuosity effects.By comparingtheoreticalpredictionswith experimentaldatafor sand,leadshot,glassheads,and snow, both the usefulness and the currentlimitationsof theoriesthat usemicroscopicpore-baseddescriptionsand their variousapproximations areexplored. 9:30 E:•. Towards a method for predicting absorption coefficient rather than measuring it in a reverberation chamber.J. Nicolas (G.A.U.S., G•nie m•canique,Universit(•de Sherbrooke,Sherbrooke,QuebecJ1K 2RI, Canada) The modelizationof the impedanceof fibrousand rigid processmaterialis discussed. For fibroussoft materialsuchas fiberglass,it will be shownthat Delany and Bazley'smodel [Appl. Acoust.3, 105-116 $9 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S9 (1970)] isquitereliablefor normalspecific impedance aswellasfor obliqueincidence. Experimental validationsaredonevia thetwo-microphone method[ submitted(1987) ] ( normalandoblique)andvia the indirect methodoftheequivalent flowresistivity (obliqueonly).Performance of lessclassical typesofsoftmateriallike felt is not wellquantifiedby DelanyandBazley,showingclearlythelimitationof thisone-parameter model. For rigid porousmaterial,the impedance behavioris generallymorecomplex.Productslike ceramicsand cement-woodfiber have beenunder investigation.Flow resistivity,porosity,and tortuosityhave beenmeasuredand introducedin Attenborough'sfour-parametermodel [J. Acoust.Soc.Am. 73, 785-799 (1983)]. Thisleavesa noteasilymeasurable parameter, theshapefactor,whichisusedasan adjustable parameter.For theserigid materials,goodagreementwill be foundbetweenpredictionand experimental dataevenfor the importantoscillations encountered forhighfrequency. Thisindicates that,withsimplesetups forthemeasurementof physicalparameter,one may be ableto calculateabsorptioncoefficient ratherthan measuringit especially fornormalincidence. However,thelinkbetween a calculated diffuse fieldabsorption coefficient and theonemeasured in a reverberation chamber is notclearat present. 9:55 E3. Optimaluseof noisecontrolfoams.J. Stuart Bolton(Ray W. HerrickLaboratories, PurdueUniversity, West Lafayette,IN 47907) Polyurethane foamis usedwith increasing frequencyin noisecontrolapplications owingto its easeof handlingandmanufacturability. Until recentlyit hasnotbeenpossible to predictitsacoustical behaviorwith theaccuracy that is possible for fiberglass, thusdiscouraging the useof foamin criticalapplications. Now theoretical modelsgoverning one-dimensional wavepropagation in noisecontrolfoamshavebecomeavailable thathavebeenshownto becapableof reproducing normalincidence absorption andtransmission measurementsaccurately. Thesetheorieshavebeenusedto predictthebehaviorof foamin severalabsorption and transmission applications. In confirmationof largelyanecdotalevidence,it hasbeenfoundthat the acoustical behaviorof foamis very sensitiveto its methodof applicationto backingsurfacesand to the way surface coatingsare applied.In addition,it hasbeenpossible to identifyoptimumtreatmentconfigurations. These studieswill be summarizedhere.Progresstowardsfull three-dimensional foammodelswill alsobe described. Suchmodelsarenecessary to allowthe prediction of obliqueincidence soundtransmission andabsorption. Finally,theoryandexperiments relatedto a newuseof foamasa vibrationdampingmaterialwill bereported. 10:20-10:25 Break 10:25 E4. Measurementof acousticimpedance of soundabsorbingmaterialsin a free field at low frequencies. Y. Champouxand J. F. Allard (MechanicalEngineeringDepartment,Universit6de Sherbrooke, Sherbrooke, Qu6becJ1K 2R 1, Canada) Thetwo-microphone technique allowstheprecise measurement ofacoustic velocity andconsequently the determination ofacoustic intensity. Usingthisprinciple, a methodhasbeenproposed tomeasure, inafreefield, theacoustic impedance ofsoundabsorbing materials. Goodresults wereobtained forfrequencies rangingfrom 700 Hz to 4 kHz [J. F. Allard et al., J. SoundVib. 114(2) (1987)]. Recently,it hasbeendemonstrated theoretically[Y. ChampouxandA.L'Esp6rance, J. Acoust.Soc.Am. Suppl.1 80, S56 (1986) ] that for low frequencies, evenif thesourceisseveralmetersfromthesample,thespherical natureof thewaveshasto be takeninto account.This paperpresents experimental evidence that confirmsthat the modelingcorrectly describes thephenomena. Furtherseveralrefinements to the experimental technique, allowingprecisemeasurementsfor frequencies as low as 250 Hz, will be demonstrated. ContributedPapers 10:50 DAMPA ceiling products.Soundabsorptionvaluesfor the textile are ES. The replacementof mineral woolin perforatedceilingsby a laminated textile. Glenn H. Frommer (DAMPA A/S, 5690 Tommerup, Denmark ) A layerof 0.2-mmtextilelaminatedontoa perforated metalbandhas beendeveloped to replaceinlaysof 12-mm34-k/m3 mineralwool in S10 J. Acoust. Soc.Am.$uppl.1, Vol.82, Fall1987 15%-20% betterat frequencies under315Hz, andgenerallyhigherat all otherfrequencies, NRC = 0.75.Thisexceptional propertyisachieved by optimizingthestreaming resistance of thetextileanditslamination.The ceilingunitsconformwith therequirements of theBritishBuildingRegulations,1976,for a class-0surfaceandtheASTM requirements for smoke density.In contrastwithotherporousmatedhals, vibrationaltestsshowno 114thMeeting: Acoustical Societyof America $10 evidence of fallout. The Danish Environmental Prize has been awarded planewaveexcitation.In orderto obtainvalidhigh-frequency data,a tube with a small crosssectionis required.But, quite often, the absorptive for the development of theseproducts. properties ofthesmallsample ofmaterialthatistestedarenotrepresentaneeringmethodhasbeendescribed[M. A. Nobile,Pro½.NOISE-CON '87,PennStateUniversity, 8-10June1987,pp.611-616]thatshows that 11:05 E6. The measurementof absorptioncoefficientand acousticimpedance using spherical waves and a transfer function method. Matthew A. Nobile (IBM AcousticsLaboratory,DepartmentC18, Building704, P.O. Box 390, Poughkeepsie, NY 12602) The normal-incidence absorption coefficient (or compleximpedance) ofanacoustical materialisusuallymeasured in a standing wavetubeusing TUESDAY MORNING, 17 NOVEMBER 1987 tive of those of the larger samp from whic it has bee tak An en largesamples of materials(or outdoorgroundsutraces)canbeevzduated usingspherical wavesin a hemi-anechoic environment. Aswiththestanding-wavetube method(per ASTM E-1050), this new methodutilizes randomnoiseexcitationand computesthe transferfunctionbetweentwo closelyspaced microphones. Theaccuracy islimitedprimarilybydiffraction from the edgesof the finitesample,whichis not takeninto account. Thispaperwill address thetheoretical aspects of theproblemandpresent somerecentexperimentalresults. TUTTLE CENTER ROOM, 9:00 A.M. TO 12:05 iP.M. SessionF. Physical AcousticsI: Nonlinear Acoustics Mark F. Hamilton, Chairman Department of Mechanical Engineering, University of Texasat Austin,Austin,Texas78712-1063 Chairman's Introductionre9:00 ContributedPapers 9:35 9:05 Fl. Nonlinear compressibilitycorrectionsto induced hydrodynamic mass: Consequences for small amplitude supersonic modes. S. Putterman,P. H. Roberts (Physicsand MathematicsDepartment, Universityof California, Los Angeles,CA 90024), and W. Fiszdon (Departmentof Fluid Mechanics, PolishAcademyof Sciences, 00-049 Warsaw, Poland) The inducedmassof an objectaccelerating throughan isotropicfluid hascontributions due to the fluid'scompressibility that are anisotropic andnonlinearin the velocity.Thiseffectcanleadto pronounced amplitudedependence intheresonant frequency ofsupersonic modes. Onesuch exampleto bepresented involvestheplasmawavethat propagates in 2-D F3. Nonlinearity parametersfor superfluidhelium. M. S. Cramer and R. Sen (Departmentof Engineering Scienceand Mechanics, Virginia PolytechnicInstituteandStateUniversity,Blacksburg, VA 240611 ) Thenonlinear propagation of second-sound in He II nearoneof the zerosofthequadratic steepening parameter associated withtheB/.4 ratio is considered. It is shownthat thecorrectdescription at thesepressures andtemperatures requires anadditionalnonlinearity parameter. Newexpressions for the nonlinearsoundandshockspeedare givenandcomparedto experiment. 9:50 arraysof ionsin He4 witha velocitythatis tentimesthe He4 sound velocity.[Work orS.P. supported byDOE andworkofP.H.R. supported by ONR. ] F4. Refractionof finite amplitudeacousticwavesat e plane fluid-fluid interface. Frederick D. Cotaras and David T. Blackstock (Applied ResearchLaboratories,The University of Texas at Austin, P.O. Box 8029, Austin, TX 78713-8029) 9:20 F2. Sound generation from a transonically moving laser. Rao J. Chandrasekharand Ilene J. Busch-Vishniac(Applied Research Laboratories,The Universityof Texasat Austin, P.O. Box 8029, Austin, TX 78712) Experimentalinvestigations of the soundgeneratedunderwaterby a movinghigh-power lasersourcehaveshownroughlya 25-ribboostin the pressure if thesourcemovesat Maeh one.A three-pronged approachto studyingthistransonicamplification phenomenon hasbeenused:an analytical examinationof the effectof includingfirst-orderentropy produc- tion termsin theequationof state,a numericalinvestigation of the effect of sourcevelocityon thesoundgenerating efficiency, andanexperimental test of the validity of superposition as a functionof sourcevelocity.Resultsshowthat nonlinearitiesand entropyproductioncannotexplainthe amplification phenomenon. Further,theacoustic efficiency isshowntobe substantiallylower for a transonicsourcethan for a subsonicsource. Hence,the transonicamplificationphenomenonis still not explainable. [Researchsupportedby ONR. ] $11 J. Acoust.Soc. Am. Suppl. 1, Vol. 82, Fall 1987 One of the oldest and most venerable laws of acoustics is Snell's law. Butdoesthislawholdfor finiteamplitudesound? Byfollowingthepathof a wavelet(point on a waveform),the followingform of Snell'slaw for a finiteamplitudewaveis derived: sin 0,/sin O, = (c• + •u, )/(c• + lg•u•), where 0, c,/g, and u are angle, small-signalsound speed,nonlinearity coefficient,and particle velocity,respectively.Subscriptsi, t, 1, and 2 denoteincidentwave,transmittedwave,fluid 1,andfluid2, respectively. SinceO,ispredicted tovaryfromwavelettowavelet,thetransmitted wave fieldcannotbetruly planar.However,by followingthepathof individual wavelets,the waveformof the transmittedsignaliscomputedat anypoint in the field. [Work supportedby ONR.] 10:05 FS. Effectsof osci!latorymotionon crystal growth.Charles Thompson andVincet Mehta (Departmentof ElectricalEngineering, Universityof Lowell, 1 UniversityAvenue,Lowell, MA 01854) 114th Meeting:AcousticalSocietyof America S11 Thisworkexploreshowthermallyinducedinstabilities drivenby the oscillatorymotionof a heatedfluidaffectthe growthof a semiconductor crystal.The mechanisms governinginteractionbetweentheinterfacetemperatureanditsgrowthwill bepresented.It will beshownthat the dynamic characteristicgrowthfront can bedescribedby a nonlineardifferential equation. Preliminary resultsof ouranalysis suggest thatmorphological defectsin bulkmaterialscanbeattributedto theaccelerated growthrate that occursin response to highunsteadytemperaturegradients.The theoryformulated suggests methods thatcouldbeusedin controlling quality of semiconductor crystals.Finally, the implicationsof this theoryin formulatingbetterprocesses for manufacturing bulk semiconductor materi- ( Khokhlov-Zabolotskaya-Kuznetsov equation).Theequation issolved numericallyfor two monochromatic sources by usinga generalization of the methodof Hamilton, Naze Tj$tta, andTjStta [ J. Acoust.Soc.Am. 78, 202-216 ( 1985) ]. Propagationcurvesand beampatternsfor the various components of the acousticfield (harmonicsand combinationfrequencies)arepresented. Specialattentionis givento thedifference frequency component. The analysisextendsto rangesbeyondthe shockformation distance.Varioussourcelevels,absorptionlengths,and downshiftratios are considered.The resultsare comparedwith that obtainedfor sources operatingat moderateintensity(quasilinear theory). alswill beoutlined.[Worksupported byAnalogDevicesProfessorship. ] 11:05 10:20 F9. Comparisonof tissuecompositionmodelsvia the measurementof the nonlinear parameter of mixtures. Erich Cart Everbach and Robert F6. Effects of boundary and fluid variations on the enhancement of E. Apfel (Departmentof MechanicalEngineering, Yale University, diffusive transport. CharlesThompson (Department of Electrical Engineering,Universityof Lowell, I UniversityAvenue,Lowell, MA 2159 Yale Station, New Haven, CT 06520) 01854) The role that bothboundaryvibrationand unsteadyfluidmotionhave in enhancingdiffusivetransport will be discussed.In particular, how acoustically inducedvorticaldisturbances interactwitha two-dimensional temperaturefield will be examined.Suchvorticaldisturbances have beenshownto bethree-dimensional [ C. Thompson,"Stabilityof a Stokes boundarylayer,"J. Acoust.Soc.Am. 81, 861 (1987) ]. Hence,parametric interactionbetweenthe flow and temperaturefield to the major mechanism for diffusionenhancement is expected.Numericalresultsdemonstratingthe degreeof augmentation will be presented.[Work supported by AnalogDevicesProfessorship and NSF. ] Recently,severalmodelshavebeenproposedfor predictingtissue composition (percentwater,protein,and fat) from physicalmeasurementsmadeonthebulk material[ Sehgaletak, UltrasoundMeal.Biol. 12, 865-874 (1986); and R. E. Apfel, J. Acoust.Soc. Am. 79, 148-152 (1986) ]. Eachmodelrequires themeasurement ofseveralbulkproperties suchasdensity,soundvelocity,andtheacousticnonlinearityparameter B/.4. Thesemodelsareintendedto aidthediagnosis andanalysisof tissue pathology. Theyalsosharewithacoustic absorption andnuclearmagnetic resonance (NMR) techniques theprospect of elucidating thecharacteristics ofintermolecular interactions in tissues. Asreportedat a previous meeting[E. C. Everbach andR. E. Apfel,J.Acoust.Soc.Am. Suppl.180, S4 (1986)], a systemto accurately measure theB/.4 valueof liquidsand semiliquid substances hasbeendeveloped. In orderto assess thesensitivity of thetissuecomposition modelsto differences in molecularstructure, B/.4 hasbeenmeasured systematically for variousproteinsandlipidsfor which acousticabsorptionand NMR data are known.To determinethe 10:35 F7. Noncollinear interaction of a tone with noise. StephenJ. Lind (Applied ResearchLaboratories,The University of Texas at Austin, Austin, TX 78713-8029) and Mark F. Hamilton (Department of MechanicalEngineering, The Universityof Texasat Austin,Austin,TX 78712-1063) validityof eachmodel,thebulkproperties of water-protein-fatmixtures are measured and the model'spredictions to the knowncomposition of eachmixturearecompared. Usingthedataobtained,theexistingmodels will be comparedand suggestions for extendingthemto moregeneral systemswill be offered.[Work supportedin part by the U.S. O•ce of NavalResearch andby the NationalInstitutesof HealththroughGrant 2-R01-GM30419-04.] The noncollinearinteractionof a tone with noiseis investigatedexperimentallyin an air-filledrectangularduct. A low-frequency bandof noisein the (0,0) modeinteractswith a high-frequencypure tone in the (1,0) mode.A quasilinear theorydeveloped byHamiltonandTenCate[J. Acoust.Soc.Am. 81, 1703-1712 (1987) ] for the noncollinearinteraction of two puretonesisgeneralized to predictthesumanddifference frequency sidebands of noisecreatedaroundthehigh-frequency tone.Whentwo pure tonesinteract, the resultingsum and differencefrequencysound oscillates in spacewith a periodicitythat dependson the frequencies and interactionangleof the primary waves.The bandof low-frequencynoiseis thusupshiftedto sidebandswhosespectralshapesare scallopedin appearance. The scallopingbecomesmore pronouncedwith range from the source,and increasingthe interactionanglereducesthe overalllevelsof the sidebands. Theory and experimentare shownto be in goodgeneral agreement.[Work supportedby NSF andby ONR.] 10:50 FS. Propagationand interactionof two collinearfinite mnplitudesound beams. Jacqueline Naze Tj•tta, SigveTj•tta (Department of Mathematics,The Universityof Bergen,5007 Bergen,Norway and Applied ResearchLaboratories,The University of Texas at Austin, Austin, TX 78713-8029), and Erlend H. Vefring (Department of Mathematics,The University of Bergen, 5007 Bergen,Norway and Departmentof MechanicalEngineering,The Universityof Texas at Austin, Austin, TX 78712-1063) The propagationand interactionof collinear,high-intensitysound beamsare investigatedwithin the schemeof the parabolicequation $12 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 11:20 FI0. Forcedradial oscillationsof singlecarlrationbubbles:A comparison of experimental and numerical studies. Co. Holt and L. A. Crum (Physical AcousticsResearchLaboratory,Departmentof Physics, Universityof Mississippi,Oxford, MS 38677) Refinements of anopticalscatteringtechniquepreviouslyreported Acoust.Soc.Am. Suppl.I 80, S24 (1986) and Suppl.I 81, S26 (1987)] havemadepossible a quantitativecomparison of theexperimentally obtainedradiusversustime (R-t) curvewith previously obtainednumerical R-t results.The response of bubbles withequilibriumradiifrom20-80 to a 22-kHz acousticfield is presentedand comparedto the resultsof previously published numericalmodels[ see,forexample,W. Lauterborn, $. Acoust.Soc.Am. 59, 283-293 (1976) andMiksisandTing, J. Acoust. Soc.Am. 81, 1331-1340(1987) ]. [ Work supportedby ONR. ] 11:35 Fll. Focal-point shifts for the reversed wave in acoustical phase conjugationexperiments:Paraxial analysisand analogy with optical holography.Philip L. Marston (Department of Physics,Washington StateUniversity,Pullman, WA 99164) A phase-conjugate mirror is onethat reverses an outgoingwavefront 114thMeeting:Acoustical Societyof America $12 sothatit propagates backtowardthesource. Recentexperiments [ Kus- 11:SO toy et aL, Soy.Phys.Acoust.31, $17-518 (1985)] indicatethat wave FI2. frontreversal canb• established through theinteraction ofa pumpwave of frequency•witha signalof frcqucncy•thatdiverges froma point source. Thenonlinear medium wheretheinteraction occurs isa thinlayer ofgasbubbles inwater.In these experiments, however, thefrequencyf• of P. O. Box 4005, Champaign, IL 61820-1305), Steve Franke (Departmentof Electricaland ComputerEngineering,Universityof Illinois, Urbana Champaign, Champaign, IL 61820), and Richard Raspet(PhysicalAcoosticsRcsearchGroup, Thc Universityof Mississippi, University,MS 38677) thereversed wavedifferedfromm= 60 kHz since f3 =f, --f• ----40 kHz andf, = 100kHz. The presentresearchis concerned with the useof a paraxialapproximation to predictthe locationof the focalpointof the reversed wave.Thenormalsofthewavefrontsoftheinteracting wavesare assumed to be nearlyperpendicular to the bubblelayer. Let z2 and z3 denotetherespective distances of the sourceandfocalpointsfrom the layer.If thepumpwaveisa planewavepropagating perpendicular to the layer,theanalysis gives% = ( •/•)z2, sothatz•z 2 whenf3-•f•. This resultisanalogous to theimagelocationin opticalholography for cases The U.S. Army ConstructionEngineeringResearchLaboratoryhas performedsimultaneous meteorological and acousticmeasurements of soundpropagation fromsmallexplosives. Categorized experimental data havebeenplacedinto groupsbasedon similarityof variationin sound speedwith height.Usingthesedatagroupsasinputto the pulsefastfield program[J. Acoust.SOC. Am. 79, 628-634 (1986) ], predictions of sound levelsare produced.The comparisonof thesepredictionsand m(asured wherethe recordingand reconstruction processare carriedout at differ- ent wavelengths[J. W. Goodman,Introductionto Fourier Optics (McGraw-Hill, SanFrancisco,1968)]. Otherapplications of thisanalogywill benoted.[Worksupported by ONR.] TUESDAY MORNING, 17 NOVEMBER Blast noi•e propagation prediction. Carolyn R. Shaffer (Departmentof Electricaland ComputerEngineering, Universityof Illinois.Urbana-Champaign, Champaign, IL 61820and USA-CERL, levels will be discussed. 1987 TUTTLE Session G. Structural Acoustics and Vibration NORTH ROOM, 9:00 TO 11:30 A.M. I: Active Vibration Control Vijay K. Varadan, Chairman ResearchCenterfor theEngineeringof Electronicand AcousticMaterials,Pennsylvania State University,UniversityPark, Pennsylvania16802 Chairman's lntraduction--9:00 Invited Papers 9:•$ GI. Active controlof soundradiationfrom platesusingforce inputs.Chris R. Fuller (Departmentof MechanicalEngineering,Virginia Tech, Biaeksburg,VA 24061) Recentlymuch interesthasarisenover the useof activecontrol methodsto reducesoundfields.The usual methodisto useanarrayof secondary acoustic sources positioned in theradiatedfield.Thistechnique has provedsomewhat successful, butonedifficultyistheexcitation of unwanted radiation(controlspillover ) that necessitates theuseofmanycontrolsources. A newtechnique isstudied in whichtheactivecontrolisapplied directlyto theradiating surface bya forceinput.Theadvantage of thisapproach is that,by reducing plate response, the radiatedsoundlevelswill be controlledglobally.The systemconsideredconsistsof a baffled clampedcircularelasticplateexcitedononesidebyanincidentacoustic planewave.Reductionoftheradiated acoustic fieldontheothersideoftheplateisachieved bya pointforceapplieddirectlytotheplatesurface. The complex amplitude ofthecontrolforceisdetermined mathematically suchthattheradiatedacoustic intensity integrated overvariousrequiredsectors isminimized.Resultsobtaineddemonstrate excellent globalreduction in the radiatedfieldlevelsfor the singlecontrolforceand the mechanisms inherentin the methodare discussed. [ Work supported by NASA Langley.] 9:30 G2.Vibrationisolafionforspaceeraft usingthepiezoelectrie polymerPVF2.SamuelW. Sirlin(JetPropulsion Laboratory,Mail Stop 198-326,CaliforniaInstituteof Technology,4800 Oak Grove Drive, Pasadena,CA 91109) Throughoutthe remainderof the currentdecadeand throughthe 1990s,the pointingrequirements of spaceborneacientificpayloadswill grow increasinglymore stringent.yet. for reasonsof costeffectiveness, the trendwill beawayfromsingle-payload-dedicated free-flyingspacecraft and towardlarge,multipayloadspace vehicles. Thebasebody structuralfrequencies of spacestation/space platformswill besignificantly lowerthan thosefor previousmissions, makingthe traditionalseparationbetweenpointingcontrollerbandwidthand structuralmodesimpossibleto maintain.Man motionand machinevibrationdisturbances will likely limit the spacestationattitudecontrolto relativelycoarselevels.In thefaceof sucha demandingdynamicenvironment, S13 J. Acoust.Soc. Am. Suppl. 1, Vol. 82, Fall 1987 114th Meeting:AcousticalSocietyof America S13 futurespacestations will hostattachedpayloads, someof whosepointingrequirements approachthoseof the free-flyingspacetelescope. Theseissuesdrivethe needfor the development of advancedpointingmount technology thatcanprovidea highdegreeof precision whilesimultaneously isolatinga payloadfroma disturbancerich hostvehicle.A finiteelementmodelof an activesoftmount basedon the piezoelectric polymer poly(vinylidenefluoride)hasbeendeveloped. The modelincludesthe geometricnonlinearities that are associatedwith the largedeflection capabilityof the softmount.This modelis put togetherwith a simplespace stationmodeland then both linear frequencydomainand time domainsimulationsare carriedout. The wideband disturbance rejection capabilities of thedesignaredemonstrated, bothin thefrequency domainand in nominaloperations. 9:55 G3. Theacousticlimit of controlof structuraldynamics. A.H. yonFlotow(Departmentof Aeronautics and Astronautics, Massachusetts Instituteof Technology, Cambridge,MA 02139) The acousticlimit of activecontrolof structuraldynamicsis investigated; the limit asthe controlbandwidthincludesa verylargenumberof naturalmodesof the structure.The pointis madethat, in thislimit, modalanalysis cannotprovide reasonably accurate models ofthestructural dynamics andthatcontroldesign with respectto modalmodelsisthenof questionable value.Alternativemodelingapproaches arereviewed.A particularwavepropagation formalism, applicable to modeling theacoustic response of networks of slender structuralmembers, isdescribed in somedetail.Controloptionsdesigned withreference to thisformalismare reviewed,and speculations asto futuredevelopments of suchcontrolare offered. 10:20 G4. Activecontrolof vibrationsusingexternallyexcitedpiezoeeramie devices.X. Q. Bao,V. K. Varadan, and V. V. Varadan (ResearchCenter for the Engineeringof Electronicand AcousticMaterials, The Pennsylvania StateUniversity,UniversityPark, PA 16802) The designandfunctionof piezoceramic devices containing eitherone(unimorph)or two (bimorph) piezoceramic disksaredescribed. Thefunction ofthesedevices istonullthetransmitted displacement fieldor boththe transmitted andreflecteddisplacement fieldswhena displacement or pressure fieldis impressed on onesideofthedevice.Sucha devicecanbeusedto activelycontrolvibrations of a givenfrequency by suitably adjustingthe voltagesthat canbe appliedto oneor boththe disks.For the unimorphdevice,fairly wideband conditions ofzerotransmission canbeachieved. Forthebimorph device, voltages tothetwodisksareindependentlycontrolledsoas to achievesimultaneously the desiredconditionsof zero reflectionand transmission. Oneor bothvoltagesources maydrainenergyfromthesystem. In eachcasetheequivalent electriccircuitcan bemodeled sothatthecurrent-voltage characteristics ofthedevicecanbecalculated andplotted.Thepiezoceramicsarecomposite in construction containing a piezoelectric phasedispersed in a nonpiezoelectric phaseso thatthedevicehastheneededmechanical andelectromechanical couplingcoefficients. ContributedPapers 10:4S G$. Active vibration control by using the wave control concept. Jiawei Lu and Malcolm J. Crocker (Department of Mechanical Engineering,AuburnUniversity,Auburn,AL 36849) model ofactive vibration control isbeing investigated experimentally. [Work supportedby SDIO/DNA, ContractNo. DNA-001-85-C-0183.] A novelactivevibrationcontrolapproachusingtheconceptof wave controlis discussed. The idea is to try to changethe characteristics of a systemand to destroythe conditionof resonance response by addinga controlforceat theboundaryof a vibratingsystem.The approachis differentfromthemodal-space controltechnique, whichisoftenusedin the vibrationcontrolof spacestructures, especially whenthesestructures are built frommaterialswith low dampingandareveryflexibledueto their largesize.For thedemonstration of thisnewwavecontrolconcept,an activevibrationcontrolmodelfor a finite lengthstringis discussed. An excitingforceis appliedsomewhere on the string,and a controlforce appliedat one end of the string is found to absorbthe vibrationalwave propagating to the endsothat no wavereflectionappearsat the endand the conditionof resonance response is destroyed.A beammodelhasalso beendeveloped. The theoreticalresultsareencouraging andbetterthan thoseobtainedby meansof themodal-space controlapproach. The beam $14 J. Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 11:00 Gf. Active control of sound fields in elastic cylinders. C. R. Fuller (Departmentof MechanicalEngineering, VirginiaTech,Blacksburg, VA 24061), R. J. Silcox,and H. C. Lester (Structural AcousticsBranch, NASA LangleyResearchCenter,Hampton, VA 2366fi) Much interest has recently arisen over the use of active noise and vibrationcontrolin variousapplicationswhereweightconsiderations are important.The resultsof research directedtowardsapplyingactivemethodsfor reducingnoisein the cabinsof propelleraircraftwill be reviewed. Two differingtechniques of implementingthe activecontrolwill be discussed.The firstconsists of placingactiveacousticsources in the interior spacein order to achievecancellation.The secondmethodrelieson active 114thMeeting: Acoustical Society ofAmerica S14 control of the transmittedacousticalenergyby vibrationalforce inputs applieddirectly to the cylinder wall. Both theoreticaland experimental Expressions of self and mutual impedanceof pistonsin an infinite, rigid, plane baffleare well known. The array is coatedwith an elastic resultsusingsimplified modelsof aircraftfuselages will bepresented and rubberskin.Plane-wave decomposition of thecompression andshearpo- the mechanisms inherent in the control action will be outlined. 11:15 G7. Interacfiooimpedanceof a systemof pistonscoatedwith an elastic skin usinga plane-wavedecomposition. Philippe Boissinot(ThomsonSintraA.S.M., ROutedesDolines,Parc d'activit•sde Valbonne,B.P. 38, for the nearfieldpressureand for the displacementfield in the sitin.This pressureis then integratedover the pistonsto obtain expressions of self andmutualradiationimpedance, whichis referredto asaveragesurface velocity.Resultsare presentedfor linear (2-D case)and circular (3-D case)radiatorsin termsof kd (dimensionless productof thewavenumber and the distancebetweenpistons).The other parametersare thickness andelasticcharacteristics of theskinandthedimension of thepiston.Asa postprocessing, thethree-dimensional directivitydiagramfor thewhole arraycanbeeasilyobtainedthroughtheplane-wave decomposition in the fluid. 06561 Valbonnc Cedex, France) TUESDAY MORNING, 17 NOVEMBER 1987 UM AUDITORIUM LOBBY AND PREFUNCTION AREA, 9:30 A.M. TO 12:00 NOON SessionH. SpeechCommunicationh Production(PosterSession} Betty H. Tuller, Chairman Centerfor ComplexSystems, FloridaAtlanticUniversity, BocaRaton,Florida33431 All posters willbedisplayed from9:30a.m.to 12:00noon.To allowcontributors theopportunity toseeother posters, contributors ofodd-numbered papers willbeattheirposters from9:30to 10:45a.m.andcontributors of even-numbered paperswill beat theirposters from 10:45a.m.to 12:00noon. ContributedPapers HI. Word-based versus syllable-basedmodels of speech production planning:Evidencefromelicitedphonological errors.StcfanieShattuckHufnagcl(Re,arch Laboratoryof Electronics, Massachusetts Institute of Technology,Cambridge,MA 02139) Earliererror-clicitation experiments haveshownthat single-segment errorsarc morelikely to be elicitedwhenthe targetsegments sharepositionin thewordonset,andlesslikelywhenthetargetsegments crossword position(e.g.,/p/and/f/are morelikelyto interactin "paradefad"and lesslikelyin "repeatfad,"eventhoughbothtargetsegments arein prestress positionin thelattercase).A separate seriesof experiments using CVC stimuli showeda differencein error patternswhen the stimulus wordsarespokenin listsversus embedded in shortphrases; stimulilike "leapnotenaplute"elicitaboutthesameproportion of errorsin initial andfinalposition, whilethephrasalversion("From theleapof thenoteto thenapof thelute") elicitsfar fewererrorsin finalposition. If thisdifference betweenword lists and phrasesextendsto other aspectsof error patterns,it might affectthe word-position/stress-position resultsfor mixedone-and two-syllablestimuli.However,resultsshowthat, when thesestimuliareembedded in phrases, theerrorpatternsremainremarkably similar.Moreover,whenspeakersare askedto generatesentences usingthewordsof the stimulus,insteadof readingor recitingthe phrases providedby theexperimenter, resultsagainsupporttheclaimthat wordonsetconsonants aremorelikely to interactin an errorthanareprestress consonants. Thesefindings suggest thatwordstructureplaysa rolein the production processing representation thatisinforcewhensingle-segment errors occur. Speech production involves theformationandrelease of constrictions at differentpointsin thevocaltract.Progress isreportedona task-dynamic computational modelof autonomous coordinationamongthe:componentsof articulatorysynergies (e.g.,lipsandjaw) that participal:e in motionsalongfunctionallydefinedtract variables(e.g., lip apertureand protrusion).For eachspeech gesture, a time-invariant dynamicalsystem (damped,second-order) is specifiedat the tract-variablelevel,and is transformed intoa yesrurallyandposturallyspecific dynamicalsystemfor thesynergycomponents. Articulatorymovement patternsemergeasimplicit consequences of the gesture-specific tract-variablecontrolstructuresandtheongoing posturalstateofthearticulators. Explicittrajectory planningis not required.Significantly,the modeledcoordinadveprocesses are exactlythe sameduringsimulations of unperturbed, mechanicallyperturbed, andcoproduced speech gestures. All simulations areimplementedusing the Haskins Laboratoriessoftware articulatory synthesizer, and are restrictedto the vocaltract'ssagittalplane,and the simplifiedarticulatorygeometryrepresented in the synthesizer.[Work supported by NIH GrantNS-13617andNSF GrantBNS-85-20709. ] H3. Source models for fricative consonants.Christine H. Shadle (Department of Electronics and ComputerScience, Universityof Southampton, Southampton SO95NH, England) Source parameters for models of fricativeconsonants werederived fromexperiments usingmechanical models having theprofileofmidsagittal x raysgivenin Fant [Acoustic Theory ofSpeech Production (Mouton, The Hague,1970)]. By positioning a pressure probeat severalpoints alongthewallof eachmodel'stract,it wasshownthatsound-generation H2. Task-dynamicmodelingof interarticulatorcoordination.Elliot characteristics of fricatives suchas/s, .•/ differfundamentally fromthose L. Saltzman, Philip E. Rubin, LouisGoldstein, and Catherine ofq,x/withregard to source location, source spectrum levela•:lowfre- P. Browman (Haskins Laboratories,270 Crown Street,New Haven, CT quencies, andtherelationship between source spectrum amplitude and meanvolumevelocity.In bothcases, soundisgenerated whenturbolent 06511) $15 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 114thMeeting: Acoustical Society ofAmerica Sl 5 air strikesrelativelyrigid boundaries.The angleof the boundarywith respect totheflowofair isthesignificant factorin thiscase.For/s, J'/, the obstacleis at fight angles,for/•, x/, the "obstacle,"the tract wall, is nearlyparallelto the flow. Implicationsfor vocaltract modelswill be discussed.[Work supportedby a Hunt Fellowshipawarded by the AcousticalSociety,and a grant awardedby the North Atlantic Treaty Organizationin 1984.] H7, Voicing distinction for fricatives: Acoustic theory and measurements.Kenneth N. Stevens (Department of Electrical Engineeringand ComputerScienceand the ResearchLaboratoryof Electronics, Speech Communication, Massachusetts Institute of Technology, Cambridge, MA02139),SheilaE. Blumstein (Department of CognitiveandLinguisticSciences, BrownUniversity,Providence, RI 02912), andLauraB. Glicksman(ResearchLaboratoryof Electronics, Speech Communication,Massachusetts Institute of Technology, Cambridge,MA 02139) H4. A twisting one-mass gloRal model. Lloyd Rice (Phonetics Laboratory,UCLA, Los Angeles,CA 90024-1543) A newmodelof vocalcordvibrationwill bepresented thatrepresents thecordbodyby a singlemassexhibitingoscillatory behavioron a horizontalaxistogetherwitha rotationalcomponent. Themodelisadequate to describethe divergingand convergingconfigurations of the glottal openingasobserved at differentphases of theglottalcycle.Characteristics of the modelare exploredundervariousglottalvibratorymodesand are comparedwith thoseof existingmodelssuchas the populartwo-mass configuration. Theoreticalanalysissuggests that simultaneous generationof vocalfoldvibrationandturbulencenoiseat a supraglottal constrictionrequires rather preciseadjustmentof the structurescontrollingthe glottis,the intraoral volume, and the supraglottalconstriction.The theory shows thattheretendsto bea reciprocalrelationbetweenvoicingamplitudeand theamplitudeof theturbulence noise.It is notsurprising, therefore,that vocal-foldvibrationdoesnot alwaysoccur throughoutthe constricted interval when a "voiced"fricativeis produced,and that variationin the noiseamplitudeisoftenobserved. Measurements of theamplitudeof voicing (expressed as the spectrumamplitudein the regionof the first harmonic) and of frication noisehave beenmade for severalfricative conson- antsspokenby severalspeakersin a numberof phoneticenvironments, includingutterancefinal and intervocalicposition,and in clusterswith othervoicedandvoiceless fricatives.Evidencefor voicingcanusuallybe foundoversomeportionof the constricted interval,particularlyin the vicinityof fricativeonsetor release.However,voicingmay be weakor absentwhen the voicedfricativeis followedin a clusterby a voiceless fricative. [Work supportedin part by Grants NS-04332and NS-15123 from the National Institutesof Health. ] HS. Bending-beam model of vocal-foldvibration. Cotinc Biekley (Departmentof Electrical Engineeringand ComputerScienceand Research Laboratory of Electronics, Room 36-521, Massachusetts Instituteof Technology,Cambridge,MA 02139} and Kevin Brown (Departmentof MechanicalEngineering,Massachusetts Institute of Technology, Cambridge, MA 02139) HS. Stresseffectson stopconsonant closureandreleasegestures. Mark K. Tiede (Haskins Laboratories,270 Crown Street,New Haven, CT 06510) Recenttheoreticaldevelopments haveshownthat a modelbasedon the theoryof bendingbeamsis the bestpredictorof the fundamental frequencies of youngchildren's speech [Bickley, Proc.11th Int. Congr. Phon.Sci.(1987)]. For childrenof age3 yearsandyounger,thevalues predictedby a spring-mass modelare too low, and the valuesof a vibrat- ing-string modelaretoohigh.In thecurrentstudy,theeffectofgrowthof thevocal-fold structureonfundamental frequency isanalyzed.Thevocalfoldstructure growsnonuniformly; forexample, a l-year-old's vocalfolds areapproximately one-fifthaslongasanadult'sfoldsbutonly 10% thinner.The bending-beam modelis an improvement overothermodelsbe- Thisstudyseeks to characterize thearticulatory kinematics of stop consonant gestures incontrasting environments, withtheaimofincorporatingtheresults intothegesture-based approach tosynthesis underdevelopmentat Haskins. TheCVCVCtokens drawnfromtheBellx-raymicrobeamcorpushavebeenexamined,representing combinations of stress, vowel,andconsonant typespokenin isolationby a femalespeaker of English. Measurements weretakenfrompeaks oftheverticalcomponent of pelletdisplacement andvelocitycorresponding to consonant closure andrelease gestures, usingthelowerlip pelletforlabialsandthetongue causeit reflectsthe structure of the vocal folds and the attachments of the root pelletfor velars.The ANOVA resultsshowthat stressaffectsconso- vocal-foldtissueto the arytenoidand thyroidcartilages, characteristic• nantgesturedurationindependently of coproduced voweltype,butinter- previously ignored. Thefundamental frequency ofthenewmodeldepends onboththebeamlike characteristics of thevocalfoldsandthespringiness of thestructure.Changes in fundamental frequency arepredicted from actssignificantly withconsonant type,possibly because velarscompete the modelasa functionof age. with coproducedvowelsfor controlof the tongue.Resultsalsoshowed different results forconsonant duration depending onwhether thegesture occurredin thefirstor finalsyllable,whichis consistent with previous studiesof prepausallengthening[D. H. Klatt, J. Phonet.3, 129-140 (1975) ]. [Work supportedby NSF Grant BNS-8520709.] H6. Estimating vocal tract shapes from x-ray microbeam data. Peter Ladefoged and Mona Lindau-Webb (Phonetics Laboratory, UCLA, LosAngeles,CA 90024-1543) X-ray microbeamtechniquesprovide data on a few points on the tongue.The possibilityof an algorithmthat will turn thesedatainto a full specification of thewholetongueshapeistested.A weakness of thesystem isthatit isdifficultto placepelletsontheposteriorhalfofthetongue.The firstgoalis thusto investigate what is the minimumnumberof pellets necessary to predicttheshapeof theanteriorpartof thetongue.Data were recordedfromsevenpoints,fourof whichwereonthefrontof thetongue, approximately15mm apart.The subjectproducedshortutterancesillustratingEnglishvowels.This procedurewasthenrepeatedwith the four tonguepelletsmovedapproximately5 mm forward.Imagescansof sus- tainedEnglishvowelsin which the subjecthad a chainon the tongue midlineconsisting of 17pellets5 mm apartwerealsorecorded.The results alloweddetermination of theextentto whichtongueshapes canbedeterminedfrom x-ray microbeampellets.[ Work supportedby NIH. ] $16 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 H9. Interactionof speaking rateandpostvocalic consonantal voicingon voweldurationin AmericanEnglish.H. S. Gopal (Calllet Center, UniversityofTexasat Dallas,1966InwoodRoad,Dallas,TX 75235) and Ann K. Syrdal (AT&T Bell Labs, IH6C-320, Naperville-Wheaton Road, Naperville, IL 60566) Voweldurationisinfluenced bothbyspeaking rateandbyvoicing of the postvocalic consonant. Our studysystematically investigates how thesetwofactorsinteract.Syllableandsegment durations weremeasured for/pVC/target syllables containing oneof eightvowelsandvoicedor voiceless stopsor fricatives. Targetsyllables wereproduced in sentence contextsat three speakingratesby four femaleand three male native AmericanEnglishspeakers. Our findingsreplicateand extendKlatt's incompressibility modelof vowelduration[D. H. Klatt, J. Acoust.Soe. Am. 54, 1102-1104( 1973) ]. The two factorsinfluencingvowelduration do not combineindependentlyin a multiplicativeor additivefashion. Instead, they interact such that vowels manifesta resistanceto further shortening. 114thMeeting:Acoustical Societyof America S16 HlO. Token-to-token variat'onof tengue-body voweltargets:Theeffect ofenarficulation. J. S. PerkellandM. H. Cohen(Research Laboratory of Electronics, Room36-543,M•ehusetts Institute of Technology, Cambridge,MA 02139) Variation ofvoweltargets fora midsagittal pointonthetongue body ofa single speaker ofAmerican English wasexamined. Thesubject pronouncedmultiple repetitionsof nonsenseutterancesof the form foV(C)Vbn/, in whichthe vowelswere/i/,/u/,/a/and theconsonant, whenpresent, wasfa/or/h/. Stress wasplacedon thesecond syllable. Movements of a singlepointonthemiddleof thetongue-body dotsum weretransduced with a newalternatingmagneticfieldmovementtransducersystem.For eachvowel,a scatterplotwasgenerated of "target" positions (x,.Vcoordinates } at themomentof minimumtangentialvelocity. In general,thesescatterplotsareelongated. Whenthevowelsin the firstandsecond syllables are the same,the scatterin targetpositionis relativelytighterand lesselongated thanwhenthevowelsare different. Whenthevowelsin thetwosyllables aredifferent,thelongaxispassing throughthescatterforonevowelisorientedin thedirectionof thetarget for the othervowel,providinga "statistically based"demonstration of contextdependence of articulatorytargets.[Work supportedby NIH Grant No. NS04332.] HI3. Stresseffectson relic gestures. RenaArens Krakow (Haskins Laboratories, 270CrownStreet,New Haven,CT 06511) In his cross-language surveyof oral and nasalvowelpatterns, Schourup [Work.Pap.Linguist. lS, 190-221(1973)] reported thatcontextualvowelnasalization ismorelikelytooccurinstressed syllables than inunstressed syllables. Thisstudy looks athowspatial andtemporal patternsof relicmovement areaffected bystress andfindsartieulatory evidencein supportof Schourup's claim.Verticalmovements of the velum were monitored whilethe(American En$1ish) speaker p?oduced utterances withdifferent stress patterns (mVbVbvsmVbVb; b'•/bVm vs bVb•rm). Velic height measured atthemidpoint ofthevowels adjacent to the nasal consonantswas significantlylower when the vowelswere stressed. Thispattern,however, primarilyreflected temporaldifferences intherelicgestures duetostress ratherthanchanges inthespatial extent ofyellolowering forthenasalconsonants. Asanexample, thevelicloweringgestures fortheinitial/m/in/m:ibab/and/mab:ib/appeared quite similarbothin theirtimingandspatialextent,butrelicraisingfromthe low positionwasinitiatedconsiderablylater in/m•bab/than/mab•b/. Thesefindingsindicatethatpartof thearticulatory distinction between syllables with differentlevelsof stressis the temporalpatternof velic movement. [Worksupported by NIH GrantsNS-13617andHD-1994.] HI4. Hll. Fundamentalfrequency and intensity control in speech. Helmet Strik and LouisBores (Institute of Phonetics,Nijmegen University,P.O. Box9103,6500HD Nijmegen,The Netherlands) A gradualdowndriftof fundamentalfrequency(F0) and intensity (IL) d.uringspeech utterances hasbeenobserved in almostall languages. Because bothF0 andIL arerelated tosubglottal pressure (P,•) andbecause P• issupposed rodecrease astheairsupplyin thelungs decreases, it hasbeenhypothesized that th/sdowndrift,usuallycalleddeclination,isa passiveprocess.In utteranceswheredeclinationis absent,like in some typesof questions, activecountermeasures wouldthenbe necessary. In thisresearch, P•, airflow,electrogiottogram, speech signal,andEMG activityof the cricothyroid,sternohyoid,and vocalismuscleshave been recorded,whilelinguistically naivesubjects sustained vowelsona fixedF0 and IL and produceda numberof sentences with differentintonation patterns. Resultsindicatethatthereisnodirectrelationbetween air sup- Generating intonational support for discourse. James RaymondDavis(AT&T Bell Laboratories, Murray Hill, NJ 07974and Massachusetts Instituteof Technology,Media TechnologyLaboratory, Cambridge,MA 02139) Properintonationcan make a spokenmessage more intelligibleby markinginformationasold or newandcommunicating the hierarchical structureof information.Use of intonationalvariationis especiallyfeasible in a systemthat synthesizes speechfrom an abstractrepresentation (insteadof from text} because the programalreadyknowsthestructureof the informationto beconveyed.This paperpresentsa schemafor assigningpitchrangeandchoosing pitchaccents.The schemadrawsuponwork on discourselB. J. Grosz and C. L. Sidncr,Cornput.Linguist.12, 175204 (1986)] andon intonationalmeaning[J. Hirschberg andJ. Pierrehumbert,Proc.24thMeet.Assoc.Cornput.Ling.,136-144(1986) ]. The modelhasbeentestedwith a programwhichgenerates spokendirections for drivingin Cambridge[J. Davis,Proc.Am. VoiceI/O Soc.(1986)]. plyinthelungs andP•, andthatP•, F0, andIL canbekeptonadesired levelwithoutactiveinvolvementof the musclesystemsthat controlinto- nationin speech. [Research supported bytheFoundagion of Linguistics, fundedby Z. W. O. ] HIS. Evaluationof speechunderstressand emotionalconditions,John H.L. HansenandMark A. Clements(Schoolof ElectricalEngineering, GeorgiaInstituteof Technology,Atlanta,GA 30332) This paperpresents resultsfrom an investigation of how stressand HI2. Respiratoryrespunseto !o• of velar resistunce.Kathleen E. Mort, Anne H. B. Putnam, and Donald W. Warren (Dental ResearchCenter, Universityof North Carolina,ChapelHill, NC 27514) The hypothesis that speech aerodynamics conformsto theprinciples of a regulating systemwastestedin ninenormaladultsubjects. The velar mechanism wasperturbed byhavingsubjects lowerthesoftpalateduring a seriesof wordsinvolvingplosiveconsonants. The pressure-flow techniquewasusedto measureoral pressures and velarresistance. Inductive emotion affectspeech characteristics withspecific application to•mprovingautomatic speech recognition. Paststudies havebeenlimitedinscope, oftenusingonlyoneortwosubjects andanalyzing onlyoneortwoparameters(typicallyinvolvingpitch). A comprehensive speech understress database hasbeenestablished at GeorgiaTechfor thepurposes uf stress research. Thedatabase ispartitioned intofivedomains, encompassing a widevarietyofstresses thatinclude: various talkingstyles( slow,fast,soft, loud,angry,clear,question, in noise }, single anddualtracking workload plethysmography wasusedto measuretidal volumesassociated with test stress inducingtasks,emotionalspeech frompsychiatric analysis session, andsubjectmotion-fear tasks.A totalof 32 speakers wasemployedto consonants. Thedataindicatethatintraoralpressure remained at appro- generate inexcess of i 6 000utterances. Thedatabase evaluation waspar- priatelevels( > 4.0 cm H20) after lossof velar resistance. Tidal volume titionedintothreeareas.Analysis wasfirstperformed on (i) speech with simulatedstress and ( ii) speech from stress inducingwarkloadtasksor speech in noise.Statistically significant parameters wereestablished, and an equivalent analysis wascarriedout over(iii) speech produced under didnotchange duringinspiration butincreased duringspeech expiration. The difference wasstatistically significant (p < 0.01). Theseresultssupportpreviousfindingsof Warren [Cleft PalateJ. 23, 251-260 ( 1986}] in cleftsubjects, Putnamet al. [J. SpeechHear. Res.29, 37-49 (1986) ] in perturbationstudies,and Warren et al. [L Acoust.Soc.Am. 67, 18281831 (1980), Folia Phoniatr. 33, 380-393 (1981 }, Folia Pboniatr. 36, 164-173(1984)] in biteblockstudiesthat intraoralpressures aremain• tainedat appropriatelevelsin thepresence of loweredairwayresistance. It is concludedthat, underthe testconditions, pressure wasregulatedby controllingrespiratory effort.Thiscompensatory response increased airflowrateandvolumewhenresistance fell. [Work supported by NIDR.] S1? J. Acoust.Soc.Am.Suppl.1, VoL82, Fall1987 actualstressand emotion.This schemewas chosensincesimulated condi- tionsallowedfor carefulcontrolof vocabulary, taskrequirements, and backgroundnoisecharacteristics.Evaluation over actual stressor emotional conditionswas usedto verify resultsestablishedunder simulated conditions. Variables employed in theevaluation include: pitch(mean, variance,highermoments, pointprocess characterization}, glottalwave- formcharacteristics, glottalsource spectrum (spectral tilt, energyconcentration),duration,intensity,formantlocationsand bandwidths,vo- 114thMeeting:Acoustical Societyof America S17 cal-tractshapingbasedon line spectralpair coefficients, and the variability ofeachofthese. Resulting categorizafious willbepresented, as wellasthestatistical significance ofcachsetofmeasurements. [Research supported byU.S. ArmyHumanEngineering Labs.] H16. The control of voice onset time and vowel duration after parancoplastiecerebellar degeneration:An acousticstudy. Susan J. Behrens, Nell Anderson, Jerome Posner, and John J. Sidtis (Department of Neurology, MemorialSloan-Kettering CancerCenter, 1275 York Avenue, New York, NY 10021) Speech production wasexamined in a groupof sixsubjects withparaneoplastic cerebellar degeneration, a neurological disorder thatproduces severe dysarthria. Subjects wererecorded repeating CVC andCVCV syllablesthatincluded thestopconsonant [p b t d k g] andthevowels[i a u]. methodcalculates a noise-to-signal (N/S) ratio,definedfromthepower spectrum,whichindicatesthe depthof valleysbetweenthe harmonic peaks.A discreteFouriertransformbasedon a continuously variable windowspanning exactlyfour fundamental periodsis usedto calculate thespectrum. A two-stage procedure isusedto determine theexactdurationof thefourfundamentalperiods.An initialestimateisobtainedusing autocorrelation in thetimedomain.A morepreciseestimateisobtainedin thefrequencydomainby minimizingtheerrorsbetweentheDFT andthe predictedspectrumof a windowedharmonicsignal.Analysisof synthesizedvoicesshowsthat the N/S ratio is sensitiveto additivenoise,jitter, and shimmer, and is insensitiveto slower (8 Hz) modulation in fundamentalfrequencyand amplitude.An analysisof pre- and post-operative runningspeech of sixpatientswithlaryngealpolypsor noduleshasconsistentlyindicatedthe successful therapeuticresultsfor all subjects.[Work supported by NIH Grant NS-13870.] Voweldurationspreceding voicedandvoiceless segments andvoiceonset time (VOT) were measured.The vowel duration measuresassessed the HI9. preservation of intrinsicdurationproperties, andthevariationof these properties in conditioning phonological contexts. The VOT measure as- dysarthric speakers. Helen Southwood and Gary Weisruer (Departmentof Communicative Disorders, Universityof Wisconsin-- sessed the coordination of two independent articulatoryevents.Results indicatedthat thecerebellarsubjects normallydifferentiated amongindividualvowels,and alteredvoweldurationas a functionof phonological Madison. Madison, WI 53706) context.In contrast,they tendedto reduceVOT for voicelesssegments, and,in thecaseof themostimpairedsubject,thevoiceless segments were alwaysperceived asbeingvoiced.AlthoughtheVOT durationswerereduced,therewasnooverlapbetweenvoicedandvoiceless VOTs for most cerebellarsubjects. Theseresultssuggest that the temporalcoordination of articulatorymaneuversrather than generaltiming of segmentaldura- tion may specificallyinvolve the cerebellum.[Work supportedby NS 17778.] Fundamental frequency contours produced by normal and Perceptual descriptions of dysarthric speech oftenincludetermsthat are meantto indexaberrancyin sentence-level fundamentalfrequency (F0) contours.Whereasthereisgeneralagreement that manydysarthric speakers produce"abnormal"F0 contours,two relevantissues that appearto be unanswered are: ( ! ) How should"normal"F0 contoursbe characterized,and (2) what is the natureof F0 contoursproducedby dysarthricspeakers. In thepresentpaper,a reportismadeon a category approach to identifying rangesof normalF0 production in youngadults andgeriatrics, andthenF0 contours ofspeakers withhypokinetic (dueto Parkinson's disease ) andspastic(congenital)dysarthriaareshown.Discussion will focuson thesuccess of the categorysystem,and the motor controlimplications of thedysarfixric F0 contours. [Worksupported by NIH.] HI7. The control of voice onset time and vowel duration after paraneoplusticcerebe!lar degeneration:Correlation with regional cerebral glucosemetabolismusingpositronemissiontomngraphy.John J. Sidtis, SusanJ. Behrens,James R. Moeiler, StephenC. Strother, Nell Anderson, Jerome Posner, and David A. Rottenberg (Department of Neurology,Memorial Sloan-KctteringCancer Center, New York, NY 10021) In a companionreport, an impairmentin voiceonset'time(VOT) controlwasdescribed for a groupof subjects with paraneoplastic cerebellar degenerationwhocouldotherwiseproduceappropriatesegmentalduration. The underlyingneurophysiological changeswere related to regionalcerebralmetabolicratesfor glucose(rCMRGIu) deterrainedby FlS-fiuorodenxyglucose/positron emissiontomography(PET). Nine cerebellarsubjectsand 18normalcontrolswerestudiedwith eyespatched whilelisteningto musicandperiodicverbalbriefings. The rCMRGIu was calculated for eachof 28 anatomically definedregionsof interest(ROI). The cerebellarsubjects couldbedifferentiated fromnormalsby theirlower globalmetabolicrate,andby theirrCMRGIu pattern.For thosecerebellar subjectswho could participatein the acousticstudy,cerebellar rCMRGlu wassignificantly belownormalrange(p < 0.05) in all but one case.Severityof VOT abnormalityappearedto be relatednot only to cerebellarrCMRGlu, but to abnormallylow corticalrCMRGlu as well. The roleof thecerebellumin maintainingtemporalcoordinationof articulatorymaneuvers appearsto involvethe coordination of corticaland subcorticalbrain regions.[Work supportedby NS17778and NS23473.] H20. Acoustic phonetic features of young children's potential homophones. George D. Allen, Diane Frome Loeb, Lori Swanson, Laurence B. Leonard, and Richard G. Schwartz (Department of AudiologyandSpeechSciences, PurdueUniversity,WestLafayette,IN 47907 ) Homophones are differentwordsthat soundalike,e.g.,"bear"and "bare." Sinceyoungchildrencan often hear differences they cannot speak,theymayproduce homophones forwordstheyknowto bedifferent.Thus,forexample,a childmayhearthat"bike"and"bite"aredifferent,butifs/he cannotyetproduce word-final consonants, s/hemayutter bothas/hal/. Two questions presentlyunderinvestigation are: ( 1) Do youngchildrenactivelyseekto includeor excludesuchpotentialhomophonesin theirearlyvocabulary? (2) Do childrenfacedwithsuchpotentialhomophones producesubtle,subphoncmic differences between them? Thispaperreports theresults ofa studyaimedatanswering thesecond of thesetwoquestions. Nine20-month-old normallydeveloping, Englishspeaking children's utterances weretaperecorded for acoustic analysis. Theseutterances wereproduced in situations in whichhomophones were likelyto resultfrom (a) reductionofinitial consonant clusters,(b) stop- pingofinitialfricatives, or (c) deletion offinalconsonants. Althoughit is hypothesized thatthemostlikelysources of subphonemic differentiation wouldbeVOT [fortype(a) and(b) pairs]andvowelduration[fortype (c) ], F0 andformantfrequencies werealsomeasured anddetailedaudi- toryphonetic analyses wereperformed. Results fromtheseanalyses will bediscussed in termsof individualandgrouppatternsof differentiation between potentially homophonous wordpairs.[Worksupported, in part, HI8. Acoustic analysis of hoarse voices in running speech. Hiroshi Muta. Thomas Baer (Haskins Laboratories, 270 Crown Street, by NIH.] New Haven, CT 06511), and Hiroyuki Fukuda (Department of Otolaryngology, KeioUniversitySchoolof Medicine,35 Shinanomachi, Tokyo 160,Japan) A methodof pitch-synchronous acousticanalysisof hoarseness requiringa voicesampleofonlyfourfundamental periodsispresented. This S18 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 H21. Cycle to cycle spectralperturbationsin the voicesof female speakers.A. Yonovitz and L. Yonovitz (Speech,Languageand LearningCenter,12841JonesRoad,Houston,TX 77070) 114thMeeting: Acoustical Societyof America S18 The glottalfunctionwasobtainedthroughtheuseof a reflectionless tube.This tubeactedas a pseudoinfinite terminationof the vocaltract. BellLaboratories, Room2D-452,600 MountainAvenue,Murray Hill, Tenfemalespeakers phonat•la neutralvowelwhiletargetmatching toa 210-Hztone.Theglottalwaveform wasdigitizedandeachcyclewaspartirionedat zerocrossings. Thisprocedure provided anaccurate determinationoftheperiodandseparate timingmeasures forthepositive andnega- In English,the alignmentof intonationpeakswith their syllablesexhibitsa greatdealofoontcxtuallygovernedvariation.Understanding this tive portionsof the glottal pulse.DiscreteFourier analysison each individualcycle indicatedchangesin the spectralcontent.The data showeda systematic changein the spectralparameters relatedto the open-close ratiofor eachcycle.Spectral perturbation analysis maybea moreinclusivemeasure of waveshape changes thanjitter or shimmer. NJ 07974) variation isof theoretical interest, andmodeling it correctly isimportant forgoodqualityintonationsynthesis. An experimental studyof thealignment of prenuelearaccentpeakswith their associated syllableswill be described. Two speakers producedrepetitions of namesof theform"Ma Lemm," "Morn LeMann," "Mamalie Lemonick," and "Mama Lemonick," with all combinationsof the four first names and three surnames. Segmental durationsandtheF0 peaklocationin thefirstnameweremeasured.Resultsshowthat althoughbothspeakingrate and prosodieoontextaffectsyllableduration,theyexertdifferentinflueaces onpeakalign- ment.Specifically, whena syllableis lengthened by a word boundary (e.g.,Ma Le Man versusMamaLetore)or stress clash(e.g.,Ma Lemm), H22. The timing of prenuclearhigh accentsin English. Kim E. A. Silverman (AT&T Bell Laboratories,Room 2C-440, 600 Mountain Avenue,Murray Hill, NJ 07974) and JanetB. Pierrehumbert(AT&T S19 J. Acoust. Soc.Am.Suppl.1,Vol.82, Fall1987 the peakfalls disproportionately earlierin the vowel.TMs seemsto be relatedto the syllable-internal durationalpatterns.It is concludedthat rulesfor generatingphoneticdetailsfrom phonologicalstructuremust access informationaboutthe upcomingprosodiccontext. 114thMeeting: Acoustical Societyof America TUESDAY AFTERNOON, 17 NOVEMBER 1987 TUTTLE CENTER ROOM, 1:00 TO 3:35 P.M. SessionI. Physical AcousticsII: Ultrasonics and Physical Effects of Sound Yves H. Berthelot, Chairman Schoolof MechanicalEngineering,GeorgiaInstituteof Technology, Atlanta, Georgia30332 Chairman's Introductionwl:00 ContributedPapers 1:05 II. Some experiments in ultrasonic pulse shaping. D. Kent Lcwis (LawrenceLivermoreNational Laboratory,NDE Section,Livermore, CA 94950)andBill D. Cook(Universityof Houston,CullenCollegeof Engineering,Houston,TX 77004) Considera pulse/echoultrasonicsystemwherethe ultrasoundis re- flectedfroma largeparallelreflector.Controlof theshapeof theoutput electricalsignalout of thissystemis attemptedby shapingtheelectronic pulseinitiallyappliedto a transducer.Analysisshowsthat transduction, both generationand reception,and propagationeachinvolvedifferentiation.Insteadof havingthegoalto bca unipolarpulse,thedesiredwave- fibrouscompositeplatesis presented.The platesare supposedto be immersedin waterandsubjected to incidentwavesat arbitraryanglesfrom the normals.For the singlelaminateplate, solutionsare presentedfor arbitraryazimuthalangles.However,for the generalmultilayeredplate case,treatmentis restrictedto the casewherethe waveis incidentalong planesof symmetryof eachindividuallamina.Reflectionand transmission coefficientsare derived, from which characteristicbehavior is identi- fied.Comparisons betweentheoryandexperiments in the formof reflected energydistributionand identificationof dispersion basedupona total transmission criterionare given. form has been chosen to bca third derivative of a Gaussian function. Initially,a Gaussian-shaped pulseis sentinto thesystem,andit isdeterminedby signalprocessing, whatmodifications of thedrivingwaveform areto bcmade,and then thisnewlyshapedpulseissentto the transducer. The outputisthenverycloseto our desiredthird derivativeof a Gaussian. Integrationof the outputsignalthreetimesyieldsa unipolarpulseof Gaussianshapewith very small kurtosis. 1:50 14. Measurementsof ultrasonicparametersin fluid-filledI•orousmetals with different grain size. Laszlo Adler, Qiang Xue (Department of WeldingEngineering,The Ohio StateUniversity,Columbus,OH 43210}, Alain Jungroan, andGerard Quentin(GroupedePhysique desSolides, Universitc Paris 7, Paris, France) 1:20 I2. Applicationof acousto-optic light modulationto the lasergeneration of ultrasound. $acek $arzynski and Yves H. Berthelot (School of MechanicalEngineering, GeorgiaInstituteof Technology, Atlanta,GA 30332} Whena laserbeampasses throughan acousto-optic Braggmodulator operatedat constantfrequency,one observes the splittingof the laser beamintoseveraldirections at anglesproportional to theinputfrequency of theBraggcell.Consequently, scanningof a laserbeamcanbeachieved Ultrasonic wavepropagation in sintered metallicsamples (immersed in fluid) wasinvestigated. Samples madeoutofbronzeandoutofstainless steelmaterialswithparticlesizesrangingfrom30--600pwereusedin this study.In orderto correlateparticlesizewith ultrasonicparameters, all sampleswerechosenwith approximately the samevolumeporosityof 30%. Fast and slowcompressional waves,surfacewavevelocities,aswell as attenuation,were measured.The frequencydependence behaviorof theseultrasonicparameters wasstudiedfrom 1-20 MHz andcorrelatedto particlesize.[Thisworkwassupported byNATO Grant0131/87 andby the NationalScienceFoundationGrant INT-8614592.] purely electronically byapplying anFM chirped signal toa Bragg modulator.A scanning rateof about4 mm//•s hasbeenobserved at a distanceof 3 m from a Braggcell modulatedbetween35 and 40 MHz in a 4-ps intervalwith a I-W cw argon-ionlasersource.Sucha scanningtechnique can be usedto enhancethe thermalgenerationof ultrasonicwavesin a sampleilluminated with a laser beam by scanningthe laser beam at a velocitythat matchesprecisely thewavevelocityin thesample.Another improvement underinvestigation is to usemultimodeopticalfibersin order not only to increasethe rangeof scanningvelocities,but also to controlpreciselythelight distributionon thesurfaceof the sample.This noncontactmethodof lasergenerationof ultrasonicwavesmay findsome applicationsin nondestructivetesting. 2:05 I5. Quantitative particle characterizationby ultrasound.Ronald A. Roy and RobertE. Apfel (Yale University,P.O. Box 2159,New Haven,CT 06520) An acousticscatteringtechniquewasdevelopedfor determiningthe compressibility anddensityof microparticles(diameterlessthan 15/zm), whichweremodeledasfluid spheres.Toneburstsof 2-/•sdurationand 30MHz center frequencywere scatteredfrom individual microparticlesas they traversedthe confocalregion of two transducers.The scattered burstswere detectedat 90' and 180' (backscattering),digitized, and storedin computermemory.By usingeitherthe Rayleighor Born ap1:35 I3. Ultrasonic wave propagation in multilayered fibrous composites. Adnan H. Nayfeh (Department of AerospaceEngineering,M.L. 70, Universityof Cincinnati,Cincinnati,OH 45221) and Dale E. Chimenti (Materials Laboratory,Wright-PattersonAir Force Base,Dayton, OH 45433) A unified analytical treatment supportedby extensiveexperimental dataof the interactionof ultrasonicwaveswith singleand multilayered S20 J. Acoust.Soc. Am. Suppl. 1, Vol. 82, Fall 1987 proximations,it waspossibleto determinethe compressibility anddensity of the particlesgivena priori knowledgeof the particlesize and host properties.It wassubsequently possible to computethe meancompressibility and densityaveragedover a populationof particles,or to display histograms of scatteredamplitudestatistics. An experimentwasrun to assessthe feasibilityof usingpolystyrenemicrospheresto calibratethe instrument.A secondstudywas performedon the buffy coat harvested from whole humanblood.Finally, Chinesehamsterovary cellswerestudied in order to seeif the instrumentcould detect hyperthermia induced membraneblistering.[Work supportedby NIH and ONP. ] 114th Meeting:AcousticalSocietyof America $20 2:20 16. Dependeneeon pulse parametersfor free radical productionby cavitation from short pulsedultrasound.J. !]. Fowlkesand L. A. Crum (NationalCenterfor PhysicalAcoustics,P.O. Box847,FraternityRow, University,MS 38677) Recentdatawill bepresented thatindicatethepresence of freeradicals generated with peakacousticpressure amplitudesandpulseshapescharacteristicof somemedicaldiagnosticdevices.The free radicalswcrc detectedvia a chcmiluminescencc reactionin an aqueoussolutioncontaining luminol using a singlephoton countingtechnique.The acoustic pressure thresholdfor lightemission wasexaminedasa functionof pulse width,dutycycle,andpulserepetitionfrequency(PRF); possible explanationsfor thesedependences basedupona simpleconceptof cavitation inceptionwill bc presented.[Work supportedby the NationalInstitutes of Health. ] 2:35 I7. Interaction of Lamb waves with surface-breaking cracks in centrifugallyeast stainlesssteel (CCSS} plates.Tran D. K. Ngoc and Kin W. Ng (Department of Physics, Georgetown University, Washington,DC 20057) Applicationof conventionalultrasonicnondestructive evaluation techniques to CCSScomponents in nuclearpowerplantshasbeenlimited principallydue to the anisotropyof CCSSmaterials.Phenomenasuchas beamskewingand distortionare directly attributableto this anisotropy and causeseveredifficultiesin defectdetectionand sizing.The present investigation is intendedto useLambwavesasthe probingmechanism to detectand characterizea surface-breaking crack residingon an interior surfacenotaccessible by the ultrasonicprobe.Experimentalinvestigation isdirectedat severaltechniques for excitinganddetectingLambwavesin thisapplication.The excitabilityof variousLamb-wavemodesin thefd (frequencytimesthickness)rangeof interestandtheeffectsof anisotropy andsurfacecurvatureon modepropagation characteristics arealsostudied. Signalsdue to mode conversionand crack diffractionare usedto developan interpretationprocedurefor crack detectionand sizing. [Work supported by EPRI undercontractRP2405-23.] 2:50 18. On a plate/surface wave mode selection criteria for ultrasonic evaluation in layered structures. A. Pilafski, $. L. Rose, and K. Balasabramaniam (Departmentof MechanicalEngineeringand Mechanics, DrexelUniversity,Philadelphia, PA 19104) The choiceof specific modeswhileutilizingplateor surfacewaves,in theultrasonic evaluation of layeredstructures, isextremely critical,especiallyif thepurpose isto tuneoursensitivity to particulartypesof imperfections. A directapproach basedontheanalysis ofeachindividualmode, for particulargeometrical configuration andfrequency content,is very time consuming.Therefore,it has been proposed[A. Pilarski, Arch. Acoust.7, 61-70 (1982) ] to usean indirectapproachby comparingthe dispersioncurvesfor differentmaterial properties,geometricalconstraints,boundaryconditions, etc.In thispaper,thistechnique hasbeen appliedto a three-layered aluminum--epoxy-aluminum adhesively bond- $21 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 ed structure.This wasachieved throughthe numericalsolutionof the characteristic equationfor sucha layeredstructure.The results'thusobtainedwereverifiedthroughexperimentally generating andstudyingthe behaviorof plate/surface wavesin specially prepared Al-epoxy-Al bonded specimens with adhesiveweaknesses. Both immersionand contact methodsareappliedin our experimental investigations. 3:05 19. Acoustically generated temperature gradients in plates. Michael Muzzerall, Anthony A. Atchley, and Thomas J. Hofier (Physics Department, Naval PostgraduateSchool, Monterey, CA 93943) The resultsof measurements of thermoacoustically generatedtemperaturegradientsin shortthin plateslocatedin a resonanttube'. are reported.The temperature gradient,resultingfroma heatflowgenerated by the acousticfield, is a functionof the acousticpressureamplitude,the Prandtlnumberof thegas,theconfigurationof theplates,andthe position of the platesin the tube,Measurements weremadefor pressureamplitudesrangingfrom approximately145-162 dB re: 20 ,uPa,in argonand helium,for singleplatesandstacksof up to fiveplateshavingseparations from approximately4 to 40 thermalpenetrationdepths,and at various positions in the tube.The resultsare comparedto predictions basedon a theoryby Wheatleyandothers[ J. WheatIcyeta/., J. Acoust.Soc.Am. 74, 153-170 (1983)]. For pressureamplitudesbelow150dB, the measurementsagreewell with theory.At higherpressureamplitudes,the agreementdiminishes, indicatingthe presence of effectsnot addressed by theory. [Work supportedby NPS Foundation.] 3:20 110. Possibilityof improvedlaser-Dopplerdetectionof soundby tracking moving phase fronts. Allan D. Pierce (School of Mechanical Engineering,GeorgiaInstituteof Technology,Atlanta, GA 30332) Laser-Doppler anemometryapplied to acousticspresentsspecial problemsbecauseof the small velocitiesof fluid flow associatedwith soundand becauseof the relatively rapid oscillations.The analysisof Hanish,attributedto GeneralElectricCompany(1967) anddemonstrated by K. J. Taylor in 1976and 1981,showssuchproblemscanbe overcomein majorpart for singlefrequencysoundusingthespectraldensityof the photomultipliertube output. Comparablesuccessis unlikely for broadbandsound;an "uncertaintyprinciple" requiresaveragingover time intervalscomparableto the reciprocalof the Doppler shift in frequency;no blackboxelectronicscan beatthis principle.Difficultiesappearinsurmountable if characteristic acousticfrequencies aremuchhigher than the Doppler shift. A possibleway of overcomingthis is to take advantageof the constancyof the soundspeedand of the omnidirectionality of incomingsound.Opticalsystems canbedesigned whichmovethe effectiveoptical scatteringvolume along nearly rectilinearpathsat the soundspeedsuchthat the laser tracksa movingphasefront. The fluid velocitysensedby the laserthusappearsto beconstantratherthan oscillating, so one has the opportunityto examinemeaningfulaveragesover sufficientlylongtime periods.[Work supportedby ONR.] 114thMeeting: Acoustical Society ofAmerica S21 TUESDAY AFTERNOON, 17 NOVEMBER 1987 UM AUDITORIUM, 1:00 TO 5:17 P.M. SessionJ. SpeechCommunicationII and Bioresponseto Vibration I: Special Focuson SpeechPerception Using Tactile Aids RebeccaEilers, Chairman Mailman Centerfor Child Development,Universityof Miami, P.O. Box 016820,Miami, Florida 33101 Chairman's Introduction--l:00 Invited Papers 1:05 Jl. Evaluations of singJe-ehanneland multichannel tactile aid• for the hearing impaired. Janet M. Weisenberger(Central Institute for the Deaf, 818 S. Euclid, Saint Louis,MO 63110) A numberof single-channel and multichanneltactile devicesfor hearing-impairedpersonshave been evaluated in ourlaboratory overthelastseveral years.Bothcommercially available andexperimental devices havebeentested,includingtheTactaidI, II, andV, Minifonator,Minivib,Queen'sUniversitytactilecocoder (both laboratoryand wearableversions),and TacticonTC-1600. Normal-hearingand hearing-impaired adults,andhearing-impaired children,haveparticipated in a varietyof trainingtasks,employing bothrecorded and live-voicestimuli. Thesetasksincludesimpledetectionof sound,environmentalsoundidentification, syllablerhythmandstresscategorization, phonemeidentification, word identification, phraseand sentence identification, connected discourse tracking,anda question-and-answer "conversation" task.In thepresent paper,theresultsfromtheseevaluations areconsidered asa whole,to permitgeneralizations aboutthekindsof informationthat canbeprovidedby single-channel, dual-channel,and multichanneldevices,with and without theadditionof lipreading.In addition,resultsarediscussed in termsof thedevelopment of"optimal"training procedures for the useof tactileaids. [Work supportedby NSF and NIH.] 1:35 J2. Speechperception throughthesenseof touchby profoundlydeafadultsandchildren.MichaelP. Lynch, RebeccaE. Eilers,and D. K. Oiler (Departmentsof Psychology and Pediatrics,Universityof Miami, MailmanCenterfor ChildDevelopment, P.O. Box016820,Miami, FL 33101) Eightcongenitally, profoundly deafchildrenandtwodeafadultsreceived unimodalvocabulary recognition training with either a multichannelelectrotactileaid or a two-channelvibrotactileaid. The children learnedto recognizevocabularies of 15-20wordswith 70%-80% accuracy.The adultslearneda 50-wordlist. Followingthis training,subjects' identification of familiarand novelwordswasassessed usingthe tactile devicesand aidedaudition.Vocabularyrecognition wastestedin eachof threeconditions:( 1) with aided audition,(2) withdualor multichannel aid,and( 3) withbothtactualandauditoryaids.Resultsindicatethat auditoryandtactualcuesweresuccessfully integratedby bothchildrenandadultsto yieldsignificantly better performance in combinedthan in singlemodalityconditions. An additionalstudyof the narrativetracking performance of oneof thedeafadultsrevealed a similarsynergistic interplayoftaction,audition,andlipreading. [Work supportedby Rita andJeromeCohen,RenataMahan, and Austinand Marta Weeks.] 2:05 J3. Lipreading with single-channelvibrotactile presentationof voice fundamental frequency. Lynne E. Bernstein,SilvioP. Eberhardt(SpeechProcessing Laboratory,Departmentof Electricaland Computer Engineering, JohnsHopkinsUniversity,Baltimore,MD 21218), and Marilyn E. Demorest(Departmentof Psychology, Universityof MarylandBaltimoreCounty,Catonsville,MD 21228) A studyevaluatedseveraltransformations ofvoicefundamentalfrequency(F0) for usebya single-channel vibrotactiledeviceto supplementlipreading.The experimentalprocedureinvolvedlipreadingstimuli from a corpusof over 1500sentences storedon a videolaserdisk[L. E. Bernsteinand S. P. Eberhardt,1986,Johns HopkinsLipreadingCorpus].There werefive experimentalconditions.Three transformations of voiceF0 were tested:two transformed F0 to a pulserate code, the third employed two sine wavesthat were amplitude modulatedasa functionofF0. Condition4 providedbothF0 and a secondstimulatorindicatinghigh-frequency speechenergy.Condition5 wasvisualalone.Subjects were 15 normal-hearingadultsin a multiplesinglesubjectdesignwith alternatingbaselineandtestsessions. Analysesof covariancefor wordscorrect,wordsper stimulussentence, and syllablesper sentenceshowedthat all subjectsimprovedin lipreading.Subjectswho receiveda singlevibratorfor voiceF0 lipreadmoresuccessfully thansubjects whodid notreceivethevibrotactile supplement or who receivedthe dual tactilestimulation. S22 J. Acoust. Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting: Acoustical Societyof America S22 2:35 J4.Tactilespeech reception usingaugmented Tadurea. C. M. Reed,W. M. Rabinowitz, andN. I. Durlach (Research Laboratory ofElectronics, Massachusetts Institute ofTechnology, Cambridge, MA 02139) Results obtained withTadurea haveseta newstandard forthetactile communication ofspeech [Reedet M., $. Acoust.Soc.Am. 77, 247-257(1985)]. Theseresults, however, areinferiorto thoseobtained in the normal auditory domain: theyareroughly comparable tothose obtained bynormal-hearing subjects listening tospeech in signal-to-noise ratiosof0-6 dB.Thegoalof thecurrent research isto augment Tadurea with supplementary tactiledisplays to improve theperception of speech segments. Among thesupplementary displays investigated werea vibrotactile display of tongue contact withthehardpalateanda vibrotactile display oftheshort-term speech spectrum. Thetongue-palate-contact display waseffective forimproving the discriminability ofpairsofconsonants difficult todistinguish through Tadurea; however, neither thisdisplay northemultichannel spectral display proved tobeeffective foraugmenting voweldiscriminability. A third approach toaugmenting Tadurea currently underinvestigation involves thetactilereception ofcuedspeech. [Work supportedby NSF. ] 3:05 JS. Temporaland spatiotemporal tactile transformsof voice fundamentalas an aid to lipreading. Arthur Boothroyal, TheresaHnath-Chisolm, andLaurieHanin(GraduateSchool, CityUniversity of New York, 33 W. 42 Street, New York, NY 10036) Lipreading scores ofnaivesubjects risedramatically whenthevisiblespeech movements aresupplemented byauditorypresentation ofvoicefundamental frequency(F0). If it canbeshownthatthetactualtransmission of F0 providessimilarresults,thenthereis the possibility of developing an effectivelipreadingaid for deaf subjects. Wehavetestedtwovibrotactile transforms ofF0. Oneistemporal.It involves single-channel stimula- tionwithaconstant-amplitude square wave,offrequency FO/2, atafixedlocus. Thesecond isspatiotemporal. It usesthesamestimulus, but locusvaries,displacement beingproportional to log(F0). Thistransformrequiresa multichannel stimulator.In psychophysical experiments, the temporaldisplayhasbeenfoundto provideroughly1/3-octresolution at thefingertip, butoverI octontheforearm.Withappropriate scaling, an eight-channel spatiotemporal displayprovidesapproximately1/3-oct resolutionon the forearm.With 16 channels,1/6 of an octaveresolution is possible. Experiments with speechstimulihaveshownthat both displays enhancethevisualperception of phoneticcontrasts that involvedurationalcues,but that themultichanneldisplayisa moreeffective lipreading supplement for contrasts involvingintonation contrasts. When sentence materialwasusedwithnaivesubjects, thetwodisplays provided onlya smallsupplement to lipread- ing.In oneexperienced lipreader, however, workingwiththemultichannel displayovera 13-week period, supplemented performance wassignificantly betterthan unsupplemented performance. The additionof the spatiotemporal transformofF0 to thevisualsignalreducedthenumberof wordrecognition errorsby 50%. A wearable versionof themultichannel displayhasbeendeveloped andiscurrentlyin usein fieldstudies.[Work supportedby NIH PPG# 17764.] 3:35-3:50 DISCUSSANT: D. KimbroughOller, University of Miarni 3:50-4:05 Break ContributedPapers 4:05 J6. A wearable multichannel vibrotactile aid for the deal A. Maynard Engebretson,FengminGong, and Michael P. O'Connell (Central Institute for the Deaf, 818 S. Euclid Avenue, St. Louis, MO 63110) to the 16-channel vocoderalgorithm,themicroprocessor-based system canbeprogrammed to implement otherprocessing algorithms usingthe existing hardware. Thesystem anditsperformance characteristics willbe described. [Work supported in partby NIH.] It hasbeendetermined that hearing-impaired and normal-hearing subjects canlearnto identifywordsusingonlya tactilevocoder[ Brooks and Frost, J. Acoust.Soc.Am. 74, 34-39 (1983) ]. Further studieshave shown thatcertainmultichannel tactilevocoders provide significant help in understanding connected discourse as an aid in lipreading[Weisenbergerand Miller, J. Acoust.Soc.Am. 82, 906-916 (1987) ]. A reviewof thisresearch isgivenin a companion paper(Weisenberger, AbstractJ1). Motivated by theseresults,a wearable, 16-channelvibrotactile aid has beendeveloped thatcanbeusedbyhearing-impaired subjects duringnormaldailyactivities. Thesystem consists of a body-worn processor thatis 3.5X 5.5X 1.5in. in sizeanda vibratorarraywornon thearm. In addition S23 J.Acoust. Sec.Am.Suppl.1,Vol.82,Fall1987 4:17 J7. Evaluation of a multichannelelectrotactiledevicefor the hearing impaired--A casestudy.Linda Kozma-Spytek and JanetM. Weisenberger (Central Institute for the Deaf, 818 S. Euclid, St. Loais, MO 63110) A multichannelelectrotactile devicefor the hearingimpairedwas evaluatedasanaidfor speech receptionandproductionwitha 12-year-old profoundlyhearing-impaired childenrolledin an oralschoolfor thedeaf. Overan 8-monthperiodthesubject received trainingin speech reception 114thMeeting: Acoustical Society ofAmerica S23 skillsin a hierarchyof tasksrangingfrom discrimination of minimal pairedwordsto connected discourse tracking.Aspectsof the subject's speech production skillswereevaluated byobtaining ratings fromagroup of teachers of thehearingimpaired,whovieweda videotape of thechild duringsyllableproductionandconnected discourse trackingwith and withoutthe device.Resultsof speechreceptiontestingindicatedthat the deviceallowedgooddiscrimination of minimalpairedwordsbasedon mannercontrasts, but poordiscrimination basedonplacecontrasts. Dur- ingtracking,lipreading-pulse-tactile aidconditions weresuperior to lipreadingalone.Resultsof the speechproductionevaluationshowedthat syllableidentification wasbetterandratingsonseveralaspects of speech production werehigherundertactileaidconditions. [Worksupported by thetactileconditionexceeded performance in theauditoryconditionfor thosestimulusitemscontaininga frequencycue. [Work supportedby #NS-17764.] 4:53 J10. Reliability of individual differences in lipreading. Marilyn E. Demorest (Department of Psychology,University of Maryland Baltimore County, Catonsville,MD 21228), Lynne E. Bernstein,and Silvio P. Eberhardt(SpeechProcessing Laboratory,Departmentof Electrical and Computer Engineering,Johns Hopkins University, Baltimore, MD 21218) 4:29 JS.Tactile receptionof fingerspelling andsignlanguage.C. M. Reed,L. A. Dclhorne, and N. I. Durlach (Research Laboratory of Electronics, Massachusetts Instituteof Technology,Cambridge,MA 02139) Previous workon tactilespeech reception by deaf-blind individuals hasfocusedon the Tadoma method [Reed et al., J. Acoust. Soc.Am. 77, 247-257 (1985) ]. Two additional methodsof tactile communication, bothofwhicharcadaptations of methods designed forthevisualsense but areusedextensively withinthe deaf-blindcommunity,arctactilefingerspellingandtactilesigning.The goalof thecurrentresearch wasto documentthe communication abilitiesof highlyexperienced deaf-blindusers of tactilefingerspelling andtactilesigning.Experiments wereconducted todeterminereception accuracy forsentence-level materialsasa function of rateof presentation for eachof thetwo methods.The resultsof these experiments will be compared to thoseobtainedfor visualreception of fingerspelling and signingas wcil as to resultsobtainedthroughthe Tadoma methodof tactile speechcommunication.[Work supportedby Evaluationof the benefitsof sensoryaidsrequiresstimulusmaterials whosepsychometric characteristics are known.Generalizability theory wasappliedto datafromtwo experiments to estimatemeasurement error arisingfrom differentmaterials,differenttalkers,and practice.Stimulus materialsconsisted of CID sentences andCV nonsense syllablesproduced by a maleand a femaletalker andstoredon videolaserdisk.In experiment I, 104normal-hearing subjects lipreadthe CID sentences in a singletest session. Results indicate that there are substantial individual differences in lipreadingperformance amongsubjects, but that therearealsosystematic differencesamongsentences and amongtalkersthat must be taken into accountin interpretingtestperformance.In experimentII, 15 normal-heatingsubjectswho participatedin an intensivetrainingprotocol involvingvibrotactilesupplements to lipreadingweregivenpre-andposttestson theCID sentences andon theCV syllables. Individualdifferences in performance on sentences werehighlystableoverthe trainingperiod, suggesting relativelyuniformimprovements due to training.Reliability overtime waslowerfor the CV syllablesand correlationsbetweenthe CV andsentence materialswereweakbothat thepretestandposttest.[Work supportedby NIH. ] NIH.] 5:05 Jll. 4:41 The use of tactile aids with deaf-blind children. Barbara Franklin (Departmentof SpecialEducation,SanFranciscoStateUniversity,San Francisco,CA 94132) J9. Stresscontrastperceptionby the hearingimpaired:Auditory, tactile, auditory-tactile.JanctRcath SchoepfiinandNancyS. MeGart (Speech and Hearing Sciences,Graduate School and University Center, City Universityof New York, New York, NY 10036) The perceptionof stresscontrastsin deafsubjects wasassessed under three conditions--auditoryonly, tactile only, and combinedauditorytactile. Test stimuli were three disyllablesgeneratedin a two-stageproccss:multiplerepetitions of eachdisyllablcwereproduced by a normally hearingspeakerusingnormalor exaggerated stresson eitherthefirstor secondsyllable;the averagedacousticvaluesfor peakamplitude,vowel duration,and peakfundamentalfrequencyfor eachdisyllablewerethen resynthcsized and manipulatedto createa stimulussetcontainingnone, one, two, and threeof the acousticcuesdenotingstress.Subjectswere askedto indicatewhetherstressoccurredonthe firstor secondsyllablefor This paperwill presentthe resultsof the firstyearof a 3-yearstudy designedto comparethe effectsof a 2-channel(Tactaid I!) and a 16channel(Tacticon)aidonexpressive andreceptive communication skills of deaf-blind children. Tbe Tactaid vibrators are worn on the wrist and the Tacticon is worn as a belt of electrical stimulators on the abdomen. Six children,rangingin agefrom22monthsto 18years,fromtheSanFranciscoBayareaareparticipating in thestudy.A totalof threecommunicative behaviors will beselected for eachchild (e.g.,numberof purposefulvocalizations).A single-subject designwill beusedtocompare theircommunicationbehaviors in threeexperimental conditions--nodevice(control), 2-channeldevice,and 16-channeldevice.Only one behaviorwill be ob- each stimulus item. Performance was above the level of chance in all three servedat a time, resultingin threeseparatesub-studies per child.This paperwill present theresults of Sub-Study # 1.Despitethetactiledefensiveness oftenexhibitedby deaf-blindchildren,thepattipantshavebeen toleratingbothdevices. [Thisresearch isbeingsupported byU.S.Dept.of conditions.In the exaggeratedstressproductionmode,performancein Educ. Grant #G008630416.] $24 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 114thMeeting:Acoustical Societyof America 024 TUESDAY AFTERNOON, 17NOVEMBER 1987UNIVERSITY LECTURE HALL, 1:15TO 3:45P.M. Session K. Psychological andPhysiological Acoustics II: Loudness andAuditoryFatigue S•ren Buus, Chairman Department ofElectrical andComputer Engineering, Northeastern University, 360Huntington Avenue, Boston, Massachusetts 02115 ContributedPapers 1:15 K1. Absolutemagnitudeestimationof loudness by adultsand children. edin accordance witha loudness-calculation programbasedonISO 532B [E. Zwicker, H. Fast, and C. Dallmayr, Acustica55, 63-67 (1984)]. Data are givenfor singletonescenteredwithin the spectrumof broad- GeorgeA.GescheiderandAmyA. Collins(DepartmentofPsychology, band-flat,low-pass,and high-passnoises.The measuredand calculated HamiltonCollege,Clinton,NY 13323) tonesassigned thesamenumberin magnitude estimation werejudgedto besubjectively equalincross-modality matching. A correction procedure wasemployed toeliminate theeffects ofidiosyncratic number usage from themagnitude estimations of loudness. Thiscorrection procedure, consistingof dividingthe loudness exponent by the line lengthexponent, loudness functionsexhibita similarpatternof loudness growth.Both measuredand calculatedloudness of the tone-noise complexes are nonmonotonicfunctionsof the overallSPL of the complex.Thus two tonenoisecombinations at nearlythesameoverallSPLcanproducemarkedly differentloudness values.Theseresultshold overa 30-dBrangefrom about73-103 dB. The resultsalsoshowthat the magnitudeof loudness depends on thespectralshapeof thenoiseandthefrequency of theadded tone.Thecloseagreement betweenthemeasured andcalculatedloudnessgrowthpatternsmeansthat basicpsychoacoustical principles governing loudness andmaskingcanaccountfor theobserved effects.[Supportedby the RehabilitationResearchand DevelopmentServiceof the VA andby produced powerfunctionexponents for loudness thatwerevirtuallyiden- the DeutscheForschungsmeinschaft. ] Twenty-three subjects performedabsolute magnitude estimation of thelengthsof a seriesof linesandtheloudness of a seriesoftonesaswellas cross-modality matching between loudness andperceived linelength.The resultssupportthehypothesis that subjects judgestimulion an absolute scale.Specifically,for 9 out of 12adultsand 9 out of 11children,linesand ticalfor magnitude estimation andcross-modality matching. Implications of the results for clinical measurement of loudnessare discussed. 2:00 1:30 K•. Deviationsfrom a power-functionnear missand their relation to loudness functions. WilliamS. Hellman(Department ofPhysics, Boston University, Boston, MA 02215)andRhonaP. Hellman(Department of Psychology, Northeastern University,Boston,MA 02115) In a previouspaper[W. S. Hellmanand R. P. Hellman,J. Acoust. Soc.Am. Suppl.1 81, S53-S54(1987)], theneural-count functionN(1) wasderivedfromthe intensity-jnd functionJ(1) throughthe relation K4. Psychometric functions for level discrimination. S6ren Buus (Communicationand Digital Signal ProcessingLaboratory,409 DA, NortheasternUniversity, Boston,MA 02115), Christine R. Mason, and Mary Florentine (Communication Research Laboratory, 133 FR, NortheasternUniversity,Boston,MA 02115) To determinethe form of psychometric functionsfor 2I, 2AFC level discrimination,ten incrementlevelswere presentedin random order withinblocksof 100trials.Stimuliwerechosen toencompass a widerange of conditions anddifference limens:eightI O-mstoneshadfrequencies of N(1)4/2= (h/2)œdl/[IJ(1) ] + a.Loudnessfunctionswerethengenera0.25, 1,8, or 14kHz and levelsof 30, 60, or 90 dB SPL; two 500-msstimuli tedby theprescription L = kiN(I) - N]. Usinga power-function near also were tested: a l-kHz missforJ(I), goodagreement betweenthemeasured andcalculatedloud- SPL.For eachcondition,at least20 blockswerepresented in mixedorder. Resultsfor threenormallistenersshowthat the sensitivityd' is nearly nessvaluesfor puretoneswasobtained. Whilea powerfunctionyieldsa simple evaluation of theintegral forN(I) •/2,theresults of manyrecent jnd studies do not easilyconformto a power-function fit. In orderto determine howthedepartures froma power-function nearmissaffectthe formoftheloudness function withinthemodel, theintegration wasperformed overthesegmented intensity jndfunctions observed forgated and continuous tones. Thecalculated loudness functions andtheirrespective inputintensity jnd functions areshownforfrequencies of 250and1000 Hz. In spiteofthesegmented, andin someinstances, nonmonotonic inten- sityjndfunctions, theresults revealthatthesmoothing actionoftheintegration overJ(I) produces loudness functions consistent withexperimen- tone at 90 dB SPL and a white noise at 63 dB proportional to AL{ = 20log[(p + Ap)/p], wherep is pressure} over the entirerangeof difference limens.Whend' is plottedagainstWeber fractions for pressure Ap/p or intensityAI/I exponents of thebest-fitting powerfunctionsdecreasewith increasingdifferencelimensand are less thanonefor largedifference limens.Theseresultsindicatethat the transformationfrom stimulusintensityto decisionvariablemay be approximatelylogarithmicand that z•L--plotted on a logarithmicscale•is an appropriaterepresentation of leveldiscriminationperformance.[Work supportedby NIH. ] tal results.[Partially supportedby the RehabilitationResearchand DevelopmentServiceof the VA. ] 2:15 1:45 KS. Investigationof a modelof loudness.P.S. Chiena> (Center for Researchin Speechand HearingSciences, City Universityof New York, K3. Overall loudness of tone-noise complexes: Measuredandcalculated. New York, NY 10036} and J. L. Hall (AT&T Bell Laboratories, Room RhonaHellman (Auditory PerceptionLaboratory,Northeastern 2D526, 600 Mountain Avenue,Murray Hill, NJ 07974) University, Boston, MA 02115) and Ebcrhard Zwicker (Institute of Electroacoustics, TechnicalUniversityof Munich, Munich, Federal Republicof Germany) Loudness measured for single-tone-noise complexes [ R. P. Hellman, J.Acoust.Soc.Am. 72,62-73 (1982)] iscompared to loudness calculat- S25 J.Acoust. Soc.Am.Suppl.1,Vol.02,Fall1907 A modelof loudnesssummationby Schroederet al. [J. Acoust.Soc. Am. 66, 1647-1652(1979) ], whichwasusedby themto provideanobjective qualitymeasureof digitalspeechcoding,is beinginvestigated.( 1) Iso-loudness contoursgeneratedby the modelare comparedto iso-loudnesscontoursgeneratedby human subjects.(2) The spreadingfunction 114thMeeting: Acoustical Society ofAmerica S25 used in the model to calculate excitation levels is converted from the Serpente, 75006Paris,France),S.Mfinikheim (InstitutffirPsychologie, critical-band domainto thefrequency domainforvariouscenterfrequencies.Theresulting curves arecompared totuningcurves measured in the cat. (3) The effectof partialmasking ontherelativeloudness of pairsof Wfirzburg,Austria),andB. Schaff(Northeastern University, Boston, puretonesisexamined. (4) Themodelisusedtopredicttheloudness of broad bandsof noisefrom the loudnessof narrow bandsof noisecentered atseveral frequencies. Studies ( 1)-(4) haverevealed strengths andshortcomings ofthemodel.a•Workdoneasaconsultant toAT&T BellLaboratories. MA02115) Measurements of inducedloudness adaptationandtemporarythresholdshift(TTS) weremadeon48 youngsubjects. Loudness adapatation of a continuous 60-dBtesttonewasinducedin therightearby an intermittent 1000-Hzinducertoneat 90 dB, presented every30 s for 20 s. The loudnessofa 1000-Hz or 1160-Hz testtone at 60 dB was measuredafter eachoccurrence of the inducerby the methodof successive magnitude estimations. Inducedadaptation causedthe loudness of the continuous toneto decrease on theaverage by 38% (the equivalent of 14dB) after 2:30 K6. Loudness adaptation in musicians andnonmusicians. C. Baruchand M.-C. Bottc(LaboratoiredePsychologie Exp6rimentale, CNRS, 28 rue Scrpente,75006Paris,France) Measurements of simpleandinducedloudness adaptationwcrcmade on 10-12 musiciansand 10-12 nonmusicians by the methodof successive magnitude estimations. Simple adaptation (SA) wasmeasured fora 500-, 1000-,or 4000-Hz tone presented monaurallyfor 3 rain at 10 dB SL. Inducedadaptationwasmeasured for a continuous 1000-Hz,60-dBSPL 120s.In a separate session, thesubject's rightearwasexposed for45rain to a 1000-Hztoneat 90 dB SPL. One minuteafter exposure,thresholds weremeasured by B6k6sy trackingfor 4 min.The maximumTTS, aver- agedacross subjects, was20.4dBat a meanfrequency of 1635Hz. The correlationbetweenmaximumTTS and the amountof inducedadapta- tionwas0.83.Thusipsilaterally induced adaptation (IIA), whichisakin totemporary loudness shift,maystemfromcochlear mechanisms justas TTS does.Also, IIA couldbecomethe basisfor an audiological testto identifythoseindividuals mostsusceptible to auditoryfatigue.[Work supported byMinist•redel'Environnement andNIH. ] right-cartone;loudness wasjudgedbefore, during,andaftereachoccurrenceof an intermittentinducertone set to five frequencies from 500- 3260Hz andpresented every30 s for 10s to thesamecar (ipsilatcrally inducedadaptation, IIA) or to theleftear (contralaterally inducedadaptation,CIA). We alsomeasured CIA with a rapidintermittcncyof every1 3:15 s for 0.5 s. The level of the inducer was 60 dB for CIA and 75 dB for IIA. Musiciansshowedlessadaptationthannonmusicians whateverthetype of adaptation:For the musicians, the maximumloudness decrease was 44% under SA, 30% under CIA, 30% under IIA; for the nonmusicians, this maximum was70% underSA, 70% underCIA, 41% under IIA, but differences werestatisticallysignificant onlyfor SA andfor CIA with the rapidintermittency. Moreover,musicians showedsignificant adaptation overa lessextendedrangeof frequencies thannonmusicians. 2:45 K9. Reductionof auditorydamagedueto intermitrenee. W. Dixon Ward (2630UniversityAvenueS.E., Minneapolis, MN 55414) Twogroups ofchinchillas weregiven dailyexposures, 5days/week for 9 weeks,to 700-2800-Hznoiseat 110 dB SPL. Each day, one group received a single 48-minexposure, theothergroupa series of40 1.2-min bursts spaced at 12-minintervals. Whereas thecontinuous single exposures produced a median permanent threshold shift(PTS)of 16dBat 1, 2, and4 kHz anddestruction of 33% oftheouterhaircells(OHCs), the intermittent exposures resulted in a PTSof only6 dBandlessthan5% destroyed OHCs.Inasmuch asthelattergroupsuffered aboutthesame cochlear damage as(andlessPTSthan)a groupgiven45daily48-min K7. A pure-tone-induced temporarythresholdshift on normalhearing and noise-inducedpermanentthresholdshift. I. M. Young and L. exposures at 102dB SPL,onecaninferthatan 8-dB"correction for D. Lowry(Department ofOtolaryngology, Jcffcrson MedicalCollcgc of dancewith present OSHA regulations governing industrial noiseexpo- ThomasJefferson University,Philadelphia,PA 19107) sure.[Researchsupported by NIH Grant 12125.] Temporarythresholdshiftafteran exposure to a puretonewasmeasuredbyautomatic audiometry onthreesubjects withnormalhearingand threesubjects with noise-induced permanent threshold shift.Subjects wereexposed to a stimulating puretonewith a frequency1500Hz throughanearphone at anintensity of 110dB SPLfor a durationof 10 min.Temporary threshold shiftwasmeasured beginning at 5 s afterthe cessation of thestimulation.In normalhearingsubjects, thegreatest shift wasobserved at thefrequency areaof 2000Hz asdemonstrated previously. In subjects withnoise-induced permanent threshold shift,temporary thresholdshiftwasdemonstrated notonlyat thefrequencyregionof 2000 Hz, but alsoat higherfrequencies with permanentthresholdshift.Post- exposure effects at higherfrequencies showed thegreaterthreshold shift forthesteady toneresulting in increased separation between normalamplitudepulsedtonetracings andmarkedlyreduced amplitude steadytone tracings. Similarpost-exposure findings wereobserved ona subject with congenital nonprogressive bilateralsymmetrical sensorineural lossof highfrequencies similarto noise-induced permanent threshold shift. Spreadof temporarythresholdshift at the higherfrequencies wasdiscussed andcomparedwith the greaterspreadof maskingeffectat the higherfrequencies. '3:00 KS. Auditoryfatigueandinducedloudness adaptation.S. Charron,M.C. Botte (Laboratoirede PsychologieExp6rimentale,CNRS, 28 rue S26 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 intermitrenee" of a 110-dBexposure is approximately correct,in accor- 3:30 KI0. Threshold recovery functions following impulse noise trauma. Roger P. Hamernik, William A. Ahroon (Auditory Research Laboratories,StateUniversityof New York, Plattsburgh,NY 12901), and James H. Patterson (Sensory Research Division, U.S. Army AeromedicalResearchLaboratory,Fort Rucker, AL 36362-5292) An analysisof the pure-tonethresholdrecoveryfunctionsobtained from 118 chinchillasexposedto high-levelimpulsenoiseshowedthat there are at leastthree distinctlydifferenttypesof thresholdrecovery functions. TypeI: theclassical recovery functionthatdeclines monotonicallywith increasing postexposure time;typeII: a delayedrecovery,i.e., for a periodas longas 6 h followingremovalfrom noise,the pure-tone thresholdremainselevatedand stablebeforea monotonicallydeclining recoveryprocess setsin; andtypeIII: the growthfunction,i.e., overa periodof at least6 h followingremovalfromthenoise,pure-tone thresholdscontinueto getworsebeforea stablemonotonically decliningrecoveryprocess begins.Frequencies characterized by a typeIII recoveryprocessshowmorePTS and sensorycell lossthan do thosecharacterizedby the type I recoveryprocess. The existence of threedistinctlydifferent audiometric recovery functions impliestheexistence of differentmechanisms of cochlear trauma. 114thMeeting: Acoustical Society ofAmerica S26 TUESDAY AFTERNOON, 17 NOVEMBER 1987 TUTTLE SOUTH ROOM, 1:30 TO 4:15 P.M. SessionL. Noise II: Analysis and Reductionof Noise Krish Krishnappa,Chairman Engineering Laboratory M-7, NationalResearch Councilof Canada,MontrealRoad,Ottawa,OntarioK1,4OR6,Canada Chairman's Introduetion•l:30 ContributedPapers 1:35 LI. Tone genemtina by impingementon piate• by suDersoniejets at v•rious over-pressures.Alan Powell (D•par•ment of Mechanical Engineering,Universityof Houston,Houston,TX 77004) Sonicjetsat "moderate"over-pressure impinging onnormal,fiat,rigid platesgenerate intenseprincipal tonesbytwodistinctmechanisms [see J. Acoust.SOc.Am. Suppl.I gl, S96(1987)]. For smallplatesd/Dg2, (d = platedlam,D = nozzlediam) a local"feedback" occursin thesobsonicregionbetweenthestand-offshockwaveand the plate,akin to that in the Hartmannair-jetgeneratorwhenoperatedat the samepressure, with •./D ___ 4(,• ----acousticwavelength).For largeplates(d/D•2) an edgetone-like acoustic feedback to thenozzleoccursthroughtheambient atmosphere externalto thejet and,•/D_2. The unstableshockwaveis excitedfrom the downstream and upstreamsides,respectively.Secondarytonesalsoarise,occurringinstead of theprincipaltone.Somehave)[/ Z)m1, and are directlyrelatedto the largeplatetones,evenfor small plates,whileothersarealsoclosely relatedtotheedgetone-like tones,but are of slightlylongerwavelength. This classification enablesa rational analysisand interpretation to be madeof data in the sparseliterature, includingthosefor supersonic jetsandjetsat "high"over-pressures, and "veryhigh"over-pressures, in whichcasethetonemechanisms appearto merge,with,•/D• 4. Panelabsorbers are very usefulin reducingthe noisepropagatedin pipesandductsin applications wherereactivemufflers cannotbeused.A panelabsorber ismadeof a rectangular panelbackedby an air gap.It is believed that this is the first time detailed studies have been made on the applications ofpanelabsorbers inpipesandducts.Oovcrning differential equations arederivedconsidering thelumpedanddistributed springmass systems. TheordinaryandpartialdLffcrential equations aresolved andthe expressions for thenaturalfrequencies andtheadmittance of thepanel absorber areobtained.Equations for theinsertionlossandthetransmissionlossof thepanelabsorber arcderived.Tendifferenttypesof alumi- numpanelabsorbers havebeendesigned, fabricated, andtestedin the Soundand VibrationLaboratoryat Auburn University.Panelabsorber parameters suchaspanelthickness, paneledgesupport(boundary) conditions,panelabsorber mounting, andpaneldimensions wereconsidered duringthe experiments. The locationof the panelabsorber alongthe lengthof thepipewasalsoexamined. Theexperimental resultsarevery encouraging. Insertion lossofthefirstpanelabsorber was2.8dB,whereas theinsertion lossofthelatestpanelabsorber withalltheimprovements is 18 dB. 2'.2O L4. Resonance conditionsin branchingstructures.David T. Raphael andM. A. F. Epstein(Departmentof Anesthesiology, Universityof ConnecticutHealth Center,Farmington,CT 06032) 1:50 In complexbranching structures, reeursive algorithms areoftenemL2. How/acousticinteractionin duetinflow.MargaretC. Quinn (The Institute of Soundand Vibration Research,SouthamptonUniversity, Southampton SO95NH, UnitedKingdom) Soundgeneration in theconvection of turbulence intoa flowintake ductisdiscussed by examination of an idealizedproblem.Idealfluidisin motionwith uniformlow, subsonic velocityaboveandparallelto a plane rigidwall.A thin,rigid,semi-infinite plateisparallelto thewallandthe fluidflowspastits leadingedge.The radiatedsoundis calculated for a turbulenteddymodeledby a weakline vortexthat is allowedto convect passively pasttheleadingedgeof theplate.Accountistakenof thecontributionto the radiatedsoundfrom the disturbance producedby the convectingvortexin theboundarylayersoneachsideof theplatebymeansof Howe's( 1981) theoryof displacement thicknessfluctuations(Proc. R. Soc.London Set. A 374, 543-568), the strengthof suchdisturbances beingfixedby a leadingedgeKutta condition.It is concluded that the predictedlevelof the radiatedsoundis substantially reducedby these boundarylayerdisturbances relativeto whentheyareneglected. The case of the convection of a frozentwo-dimensional gustis alsoconsidered. Examination of theanalogous problemof plane-wave radiationfromthe duct showsthat the presence of displacement. wavesenhancedboth the far-fieldintensityin the ambientfluid and the reflectedfieldwithin the duet.[Work supported by ARE Teddington, UK.] 2.'O5 L3. Further studieson pimel absorbers.B. S. Sridhara and Malcolm J. Crocker (Department of Mechanical Engineering, Auburn University, Auburn, AL 36849) S27 J.Acoust. Sec.Am.Suppl.1,Vol.82,Fall1907 ployed tocalculate impedances atsuccessive nodes. Fora rigidsymmetric N th-orderbranching networkterminating in 2N closed branches, a techniqueispresented whereby theopen-ended inputimpedance, anditscorresponding resonance condition, canbeexplicitly formulated intermsof thebranchlengthsandcross-sectional areas,theangularfrequency, and theintegralharmonic number.In thelow-frequency range,theresonance condition can be modified so as to facilitate volume estimation of the branching structures. Experimental studies withrigidseeond-order symmetricandasymmetric branching structures arediscussed. [WorksupportedbytheUniversity of Connecticut Research Foundation. ] 2:35 IS. Surfaceroughness effectson flow noise.OeorgeH. Christoph (Martin Marietta Laboratories,1450S. Rolling Road, Baltimore,MD 21227) TheCorcos wallpressure modelwasmodified to incorporate surface roughness effects. Thelongitudinal cross-spectral density partoftheCotcosmodelwasmodeledfrom roughwall datatakenfrom theliterature. The wall pressure frequency spectrum wasnondimensionalized by the wallshearstress andboundarylayerdisplacement thickness andshownto fit bothsmoothandroughwalldata.Theboundarylayerparameters were obtainedfrom a finite-difference boundarylayer codedevelopedby the authorto accountfor surfaceroughness effectsby thediscreteroughness approach[O. H. Christophand R. H. Pletcher,AIAA L 21, 509515(1983) ]. The blockage of theroughness height,shapeandspacingis explicitly accounted forinthisapproach. Flownoise wascalculated fora varietyofroughness elementheights andspacings in airandin water.The 114thMeeting: Acoustical Society ofAmerica $27 calculations clearlyshowthatelementspacing aswellasheightisimportant and that the viscoussublayerheight,not the boundarylayer thickhess,determines theinfluenceof roughness onflownoise.Thusroughness is typicallymoreimportantin waterthanin air because of the thinner viscous sublayer.Also,theeffectof transducer sizeandviscoelastic layer thicknessare shownfor roughsurfaces. 2:50-3:00 Break 3:00 L6. Helicopter rotor speedeffectson farfield acousticlevels.Arnold W. Mueller, Otis S. Childtess,Jr. (NASA Langley ResearchCenter, M.S. 460, Hampton, VA 23665), and Mark Hardesty (McDonnell DouglasHelicopterCompany,Mesa,AZ 85201) The designof a helicopteris basedon an understanding of manyparametersand their interactions.For example,in the designstageof a helicopter,the weight,engine,and rotor speedmustbeconsidered along with the rotor geometrywhenconsidering helicopteroperations.However,the relationshipbetweenthe noiseradiatedfrom the helicopterand theseparameters is notwellunderstood, withonlya limitedsetof model and full-scalefield test data to study.In general,thesedata have shown thatreducedrotorspeeds resultin reducedfarfieldnoiselevels.Thispaper will reviewthestatusof a recenthelicopternoiseresearch projectdesigned toprovideflightexperimental datatobeusedforfurtherunderstanding of helicopterrotor speedeffectson farfieldacousticlevels.Preliminaryresultswill be presentedrelative to testsconductedwith a McDonnell DouglasHelicopterCompanymodel500Ehelicopteroperatingwith the rotorspeedasthecontrolvariableovertherangefrom 103%N2 to 75% N2 andthe forwardspeedmaintainedat a constantvalueof 80 knots. 3:15 L7. Fan noise and unsteady rotor forces. Wen-Shyang Chiu, Gerald C. Lauehle, and D. E. Thompson (Graduate Program in Acoustics, Applied Research Lab, PennsylvaniaState University, Box 30, State College,PA 16804) The noiseradiatedby a subsonic,axial-flowfan at its rotationalfrequencyandharmonicsisrelatedto thenonsteady forcefieldcreatedat the rotor blade/fluidinterface.This forcefield is highly dependenton the time-invariantflow distortionsthat enter the fan. In this basicstudy,a typicalcoolingfan usedin theelectronicandcomputerindustrywasinstrumented withanunsteady axialforcesensor. Its outputisproportional to the total unsteadyaxial forcecreatedby the rotor. The inflow field of thefan wassystematically distortedby placinga smallcylinderat various positionsin theinlet plane.The nonuniform,three-dimensional flowfield enteringtherotorwasmeasured by traversing a miniaturefive-holepressureprobe.The total pressureoutputsfromthisprobecanberelatedto the axial,tangential,andradialvelocityvectors.Fourierdecomposition of the inflowvelocitydata is coupledwith analysisto giveinformationon the unsteady rotorforceharmoniccontent.Theon-axissoundpressure levels weremeasured andcomparedto coherentoutputpowerspectrainvolving the unsteadyforcesensorandthe microphone. Very goodcoherence at the discretetonesis observed.[Work supportedby IBM Corp.] in-ductsilencer.Pressuredrop, fouling,and other problemsfrequently associated with silencers canalsobeavoided.Previousattemptsat modi- fyingan existingcentrifugalfan to controlnoisehaveoftenmet with limitedsuccess. To reducethe costof controllingbladepassage tonesin the community,modificationof the induceddraft (ID) fansat a power plantwasevaluated. Modeltestingwasconducted for ten differentdesignsof thecutoffbarfor thecentrifugal fan.Thenoisereductions andfan efficiencies varied widely for six of the ten cutoffbars tested.The best designgavebetternoisereductions thanexpected andhigherefficiencies at usualsettings ofthevariableinletvanes.The bestcutoffbardesignwas installed (full scale) on one of the ID fans and fieldstested. The noise reductions on the full-scale fan tests were better than obtained for the scale modeltests.Resultsof both the modeland full-scaletestsare presented, compared,and discussed. 3:45 L9. Frequencydomainanalysisof dipolesourcestrengthand correlation area. David M. Yeager (IBM AcousticsLaboratory,P.O. Box 390, C18/704, Poughkeepsie, NY 12602) The generation andradiationof noisefroma rigidsurfaceimmersedin a turbulentfiowfieldmaybe characterized by the strengthof the dipolelike sourcesassociatedwith the normal surfacepressuresacting on the fluidat thesurface, andbytheregiS)ns overwhichthese sources arepartially correlated.New frequency-domain expressions have beendevelopedfor the dipolesourcestrengthper unit area, basedon mean-square pressure andacousticintensity,and for the correlationarea.The equationsaresimilarto time-domain expressions developed earlierIT. E. Sialdon,J. Acoust.Soc.Am. 53, 619-633 (1973) ]. The equationsshowthat the correlationareacanbe thoughtof asa complexradiationefficiency that is normalizedby its maximum value, the total area of the surface immersedin turbulence.Both the correlationarea and dipole source strengthper unit area reflectthe noisegenerationcharacteristics of the localturbulenceregionsandmutualinteractions betweenregions.Results of experiments will bepresented in whicha singleblade,removedfroma rotatingcentrifugalair movingdevice,wasimmersedin a turbulentflow field. 4:00 LI0. Scalemodelof an activenoiseredactionsystem.O. L. Angevine (AngerincAcousticalConsultants, Inc., P.O. Box725,EastAurora,NY 14052) 3:30 LB. Model and full-scale testing of a modified cutoff bar for a centrifugal fan. Frank H. Brittain (Bechtel National, Inc., San Francisco, CA 94105) and David A. Nordstrand (Detroit Edison Co., Detroit, MI 84226) Controllingcentrifugalfan noiseby modifyingthedesignof the fan is very appealingbecausethe costcouldbe considerably lowerthan for an S28 d. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 Scalemodelinghasbeenusedoftenfor predictingacousticalproperties of architectural structures such as music halls. In this experiment, it is an activenoiseattenuatingsystemthat ismodeled.By usinga scalefactor of 8 or 16, testscanbeconductedon a tabletop simulatingthoseof a fullscalesystemat a distanceof 24 m. Sinceactiveattenuationismostsuccessful at frequencies belowabout500Hz, thescalefrequencies remainin the audiblespectrum,and measurementdoesnot requirespecialequipment. [Work supportedby EPRI and NMPC. ] 114thMeeting:AcousticalSocietyof America S28 TUESDAY AFTERNOON, 17 NOVEMBER 1987 UMS 4 AND 5, 1:30TO 5:00P.M. Session M. Underwater Acoustics II: ArcticAcoustics II (Prtcis-Poster Session) Robert Francois,Chairman Applied Physics Laboratory, University of Washington, Seattle,Washington 98195 ContributedPapers Followingpresentation of thepr•cis,posters will beondisplayuntil5:00p.m. 1:30 M1. Ambientnoi•ein BaffinBay.Joseph B. FarrellandGarryJ. Heard (Defence Research Establishment Atlantic,P.O.Box1012,Dartmouth, The seaiceis modeledasa transversely isotropicbrine-saturated porousmedium,in accordance with the modelproposed by Schwarzand Weeks (1977). The reflection and refraction of waves at the water-ice This paperpresents the resultsof ambientnoisemeasurements taken in BafflnBay in August-September of 1986.Noisemeasurements were interfacearestudied.Thenumericalresultsobtainedin thisstudyindicate thattheporosityof theicehasa noticeable effectonthewavereflectivity. Its effect,however,becomes smallwhenthegrazingangleof theincident wavebecomes small.The existence of a thin andporousiceskeletonlayer takenin conditions rangingfromopenwaterto iceberg-infested water on the icemediumfurtherreducesthe reflectivity.The reductionof reflec- neartheedgeof packice.Differences in thecharacterof ambientnoiseat tivitybecomes moredistinctasthefrequency of theincidentwaveishigh. Numericalresultspointoutthat,in additionto theporosity,theicelayer thickness is the mostdominantpropertyinfluencing the undericewave reflectivity.The calculatedresultsshowa qualitativeagreementwith the Nova Scotia B2Y 3Z7, Canada) theexperimental siteswillbenotedanddiscussed. Thefrequency range investigated in thisworkwas10-1000Hz. Comparisons will bemadewith noisemeasurements takenby DefenceResearchEstablishment Atlantic (DREA) in BailinBay in the 1970s. in situ experimentalresultsby Yang and Votaw ( 1981). 1:34 1:42 M2. Super-resonantscatter under ice sheets and in full spaces. Ivan Tolstoy (Knockvennie,Castle Douglas, SW Scotland) and M4. A modal approachto Arctic noisemodelingwith implied vertical AlexandraTolstoy (Naval ResearchLaboratory,Washington,DC coherence. R.B. Williams(NavalOceanSystemsCenter, SanDiego,CA 20375) 92152) Systems of monapole scatterers interacting throughmultiplescatter may developnormal modes,i.e., true resonantbehaviorat certain frequencies(the equivalentsourcecoefficients of eachscattererhavereal poles).Thisoccurswhenthesinglescattercrosssections arelargeenough to givesignificantinteraction.Thishappensfor highlytunedscatterers, e.g., gas-filledballoons,bubbles,and thin shellsin water, insonifiedat frequencies neartheircharacteristic "bubble"frequencies. Thiskind of super-resonance (SR) occurswhen the systemis at or near the interface with an elasticmedium,e.g.,a thin elasticplate [I. Tolstoy,J. Acoust. Sac.Am. 80, 282-294 (1986)]. Energytransportbetweenscatterers is thenmediated byboundary modes excitedbythescatterers actingaspoint sources(e.g., the flexurewavesof a thin plate). The caseof two and three bubblesystems at an ice-wateror lucite-waterinterfaceisexplored,with particularreference to theparameters ka, kl (k istheacousticwavenum- ber,a thescatterer radius,l thespacing between scatterers) andtheplate elasticconstants. For a pair (doublet)the pressureamplificationre: the free-fieldvalueis _•2 X 103for kl valuesassmallas0.5, with a directionality factorof 30: This is a passive,highlytuned,high magnification, stronglydirectionalcompactarray ( •_A / 12size).A 30-Hz systemwould Analyticformulasof the modalattenuation coefficients appropriate for deepArctic under-iceconditions, and modalenergypartitioningof ambientnoisesources areproposed. Thesefunctionscanbeused,together with variousassumed spatialand temporaldistributionfunctionsto calculate various statistical ambient noisefunctions, such as vertical coher- ence.Verticalnoisedata between15 and 50 Hz havebeenanalyzedfor modelcomparison,usinga bandwidthof about 5 Hz. Thesedata show thattheverticalcoherence usuallyhasa sinx/x dependence, withthefirst zero crossingat one wavelength.Sometimes, however,zero crossing lengthsof morethan two wavelengths are observed.Longercorrelation lengthsare consistent with muchof the noiseenergyradiatingfrom a smallregion,althoughotherinterpretations are possible. Initial simple modelcalculations showagreement with thedata.A strongnearbysource of noisewill causethecorrelationto lengthen,whileseveralmoredistant sources supporttheonewavelength zerocrossing result.Initial sensitivity calculations, however,showthat the verticalcoherence functionby itself is not a goodvalidatorof the modelfunctions.Other statisticswill be soughtto furtherexaminemodelvalidity. [Work supportedby NORDA and NOSC 6.2 Arctic programs.] consist of twoballonsof 10-cmradiusnearan iceboundary, witha discretesetof permissible I valuesrangingfromabout5-20 m. Optimum spacings andmagnifications dependupontheicethickness( _ 5 m). In an acoustic full space,energyexchange betweenscatterers is mediatedless effectively by volumewaves,and the scattercoefficient polesvanish: Bona-fideSR doesnot takeplace.Nevertheless, strongpeaksmayoccur, givingmagnifications _ 103.Thisquasiresonant behavior isdiscussed for simpletwo-, three-,four-,andsix-element planeandpolybedralconfigu- rations.However,trueSRbehavior isreintroduced in full spaces by partiallyblocking selected pathways between scatterers. [Worksupported by ONR. ] 1:38 M3. A studyof reflectionandrefractionof wavesat the interfaceof water and sea ice. C. H. Yew and X. Weng (Departmentof Aerospace Engineering and EngineeringMechanics,Universityof Texasat Austin, Austin, TX 78712) S29 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 1:46 MS. Under-ice scattering mechanisms.R. H. Mellen (PSI Marine Sciences, New London, CT 06320) and P. S. Seheifele (Naval UnderwaterSystemsCenter,New London,CT 06320) Underwater acousticpropagationin the Artic Ocean is characterized by a refractivesurfacechannelwith a roughboundary.Scatteringestimatesbasedon free-surface theoryhaveprovenlow by morethan a factor of 2 in forward-scatterlossand low by 20 dB or more in backscatter strength.Failure to accountfor eitherattenuationor backscattering indicatesthatimpedance effects mustbeinvolved.The icelayerismodeledas a uniformelasticsolid.Perturbationanalysisshowsscatteringfrom slopes aswellasdisplacements. Displacements producepistonlikereradiationin addition to the usualvertical-dipoletype associatedwith free surfaces. However,the "rocking"horizontaldipoleproducedby slopesis evidently the dominantmechanismat the lower frequenciesand estimatesof both 114thMeeting:Acoustical Societyof America $29 attenuationand backscatter strengthshowreasonable agreementwith experimental datawhenincluded.[Worksupported by NUSC/NL.] or nosnowcover.Sincetheonlyicesurfaces directlyexposed to air in the vicinityof theexperiment werethe verticalsidesof largeblocksof icein pressure ridges,it is hypothesized that the crackscouldhaveoriginated there. 1:50 M6. Sound ubsorl•tionin Arctic regions.R. H. Mellen (PSI Marine Sciences, NewLondon,CT06320),P.S. Scheifele, andD. G. Browning (Naval UnderwaterSystemsCenter,New London,CT 06320) Soundabsorption in seawater involves relaxations of magnesium sulfate,boricacid,and magnesiumcarbonate.Absorptionvarieswith both regionand depth, due mainly to the pH dependence of the boric acid relaxation.The nominalseawaterpHrangeis7.7-8.3 andmagnitudes can varyovernearlya factorof4 at lowfrequencies. A globalmodelforWorld Oceanpredictionhasbeendeveloped, employingcontourchartsof thepH factorat severaldepths.Profilesfor integration of lossalongraypathsare generatedby algorithm.ThepH rangein the Arctic isnominally8.0-8.3 andsimilarcontourchartsareprovidedfor thepH factorin theseregions. Althoughlong-rangepropagationis generallylimitedby under-icescattering,absorptioncan be the dominantlossmechanism underice-free conditionsand for rangeslessthan 50 km at sonarfrequencies with an ice cover.Predictedabsorptionspectraare presentedand comparedwith the limiteddataavailable.[Work supported by NUSC/NL.] 1:54 M7. A bistatic, high-frequency,under-ice, acoustic scattering model. Garner C. Bishop (Bldg. 679, Code 8211, Naval Underwater Systems Center, Newport, RI 02841) A modelhas beendevelopedto evaluatethe scatterproducedby a high-frequency acousticpulseoriginatingfrom an arbitrarily located source,incidentontheunder-icesurfacecharacteristic of packiceregions of the interiorArctic and detectedby an arbitrarilylocatedreceiver.Measured,two-dimensional, under-ice,acousticprofiledata and severalem- piricalresultsthat relatevariousgeometric parameters of thelargescale 2:02 M9. An experimentnlstudyof the coherentunder-icereflectivityof sound in the GreenlandSeaMarginal Ice Zone (MIZ). PatriciaL. Gruberand RonaldL. Dicus (Code5120,Naval ResearchLaboratory,Washington DC 20375) The coherentcomponent of acousticunder-ieereflectivitywasinvestigatedasa functionof frequency andgrazinganglein theMIZ. Measured reflectivitywascomparedwith scatteringtheorypredictionsto test the hypothesis of ridgescatteringdominancein the MIZ. Explosivesource acousticsignalswerereceivedona verticalarray,duringtheMarginalIce ZoneExperiment(MIZEX 84), anddeconvolved to separatethe direct and ice-reflected arrivals.Ice-reflectedsignalsreceivedon individualhydrophonesat differentrangesand depths,and on differentdays,were alignedin timeandcoherently ensemble averaged withina moving10-deg grazinganglewindow.The resultingreflectivitywasfoundto decrease with increasingfrequencyfrom 64-256 Hz and with increasinggrazing angle from 12-35 deg. Comparisonwas made with predictionsfrom smoothelasticplatetheory,perturbationtheoryfor a roughpressurereleasesurface,perturbationtheoryfor a roughelasticsurface,Twersky theoryfor hemispherical bosses on a plane,andTwerskytheoryfor infinite,semielliptical cylindersona plane.The Twerskytheoryfor semielliptical cylindersgave a significantlybetter fit to the data than the other theories.This suggests that, althoughtheunder-icetopographymaybea gooddeal morecomplicated,scatteringmay, nevertheless, be dominated by remnantsof elongated pressure ridgekeels.Remotesensing data,taken duringtheexperiment,showedthat theicecoverwasprimarilycomposed of smallfloes( < 200 m in diameter).Laserice-surface heightmeasure- mentswereanalyzedto obtainthepressure-ridge sail-height distribution. The Twerskytheorymodelinggavepredictionswithin 1 dB of the measuredmeanreflectivitywhena ridgekeel-to-sailratioof 6.5 wasassumed. under-icerelieffeatures,e.g.,icekeelsare usedto constructa three-dimensional,bimodal,under-icesurfacemodelconsistingof first-yearice keels andslopingflat iceregions.A first-yearkeelis modeledasan ensemble of randomlyorientediceblockson a planarsurfaceinclinedat someslope anglewith respectto a horizontalplaneat sealevel.Iceblocksaremodeled aslayered,viscoelastic, rectangularsolids.A regionof flat ice is modeled as a smoothplanar surfacewhoseslopeangleis lessthan somearbitrary minimumkeel slopeangle.The Helmholtz-Kirchhoffintegraland the Kirchhoffapproximation are usedto evaluatethe scatteredfield.Time is partitionedinto binsand then,beginningat the bin corresponding to the timethe scatteredfieldof eachscatteringfacetarrivesat the receiver,the scatteredfieldis addedcoherentlyin all binsspanningthe temporalduration of the incidentpulse.The resultingscatteredfield time seriesis used to calculatethe corresponding reverberation time series.Modeledreverberationlevelsarecomparedwith measured levelsfor severalArctic sites and modestagreementis obtained. Observation of thermal ice.crnekinR MIO. Attenuation of sound through laboratory grown saline ice. K. C. Jezek (U.S. Army Cold Regions Research and Engineering Laboratory,and Thayer Schoolof Engineering,Dartmouth College, Hanover, NH 03755), T. K. Stanton (Department of Geology and Geophysics,Universityof Wisconsin,1215 W. Dayton, Madison,WI 53706), and A.J. Gow (U.S.A. Cold RegionsResearchand Engineering Laboratory, Hanover, NH 03755) Attenuationof acousticwavestransmittedthrough salineice was studiedin situasiceformedin a large,outdoorpool.Amplitudeof sonar pings (50-800 kHz) transmittednormal to the ice sheetsurfacewere measuredafter passingthrough ice about 18 cm thick and having an averagesalinityof about4 paRRs per thousand.Columnarcrystalsin the body of the ice sheetand a fine, dendriticstructurenear the baseof the ice sheet(dendriteswere about0.5 mm thick) characterizethe ice as being 1:58 MS. 2:06 nearlyidenticalto young,uncleformed seaice foundnaturallycovering in the Arctic Basin. the polaroceans.Additionalmeasurements wereperformedon newly P. Zakarauskas (Defense Research Establishment Pacific, FMO, Victoria, British Columbia V0S lB0, Canada) grownsalineiceabout3 cm thick.Measuredattenuations throughthe 18cm-thickicerangedfromabout20 dB at 50 kHz to 80 dB at 420 kHz. For It is now well knownthat thermalice-crackingis a major contributor to under-iceambientnoisein certainregionsof shore-fast iceduringwinter andspring[A. R. Milne, J. Geophys.Res.77, 2177-2192 (1972) ]. A reportisgivenhereof an observation of thermalicecrackingmadeby the DefenseResearchEstablishmentPacific over polar pack ice during the springof 1987.The numberof events,ascountedby a time-domainkurtosismethod,is showedto followthe temperaturechangesof the air very closely. Thermal ice-crackingwas the dominant ambient noise source duringtheperiodsof rapidair cooling.For thermalice-cracking to occur, threeconditionsmustbemet:low temperature,low ice salinity,and little paredto about40 dB for the 18-cm-thickice). That this represents an absorption of theacousticwaveby theiceisevidenced bytwo otherobservations:first, a low reflectioncoefficientmeasuredon newly grown ice (about0.04 at 188kHz); second,measurements of scatteredenergyback $30 J. Acoust.Soc. Am.Suppl.1, Vol.82, Fall 1987 the 3-cm-thick ice, total attenuation at 188 kHz was about 12 dB (com- into the transmitting region showedlittle redistribution of energy into nonspecular directions.Basedontheobserved decrease ofattenuationper unit thicknessof ice, significantabsorptionis attributedto the lower, porousand permeableportionsof the ice. Presumablythe importantabsorptionmechanisms are associated with the pureicedendritesand alternatingplanesof brinepockets. 114thMeeting:AcousticalSocietyof America $30 2:22 2:10 MII. Elements of Arctic surface scatter: Part III, the head wave. MI4. Sea ice reflectioncoefficientestimatesat low frequency.George HermanMedwin,Ken J. Reitzel, =! andMichaelJ. Browne(Physics R. Giellisand T. C. Yang (Naval ResearchLaboratory,Washington, Department,NavalPostgraduate School,Monterey,CA 93943) DC 20375-5000) Laboratoryscalemodelstudieshavethe virtueof revealingthe detailedbehaviorof the elementsof scatterfrom a complexice field.Previousmodelstudieshavedemonstrated thequasidipole radiationpattern of low-frequency scatterfroman openlead.It hasalsobeenshownthat leadsand ridgesbackscatterprincipallyfrom a one wavelengthextent neartheleast-timeinterceptof theradiationfroma pointsource.At this time, evidenceis presentedthat mode conversiongeneratesa compressionalwavethatpropagates parallelto thesurfaceof theice.Thismodeis driven,andwill radiateintothewater,at thecriticalangle.The reradiated sound,anexampleof a headwave•causes a significant precursorto appear beforethe specularreflectionfrom the ice canopy.In agreementwith Resultsof a seaicereflectivityexperimentconducted at theFRAM IV icecampwill bepresented. A verticalarraywith27 receivers at depths rangingfrom 30-960 m wasemployed.Implosivesourcesat various depths(30-120m) weresetoffat a rangeof 1145m fromthereceivers. Analysisof directandreflectedpulsesat the receivers resultedin reflectioncoedticient estimates at frequencies in the 50 Hz to I kHz rangefor reflection angles between 6' and60'. Theresultswill becompared with previous data[T. C. YangandC. W. Votaw,1.Acoust. Soc.Am.70,841851 ( 1981) ] and modelcalculations. theory, themeasured headwaveamplitude isproportional to XL -•t2 wherek is the wavenumber in the water,x is the horizontal range,andL isthepathlengthin theice.[ WorksuppoRed by theOfficeof NavalResearch.] *•OceanAcoustics Associates, 4021Sunridge Rd., Pebble Beach, CA 93953. 2:26 MI5. Two numericalmodels•one simplistic,the othersophisticated-applied to under-ice acoustic propagationand scattering. Stanley A. Chin-Bing (Naval Ocean Researchand DevelopmentActivity, Numerical Modeling Division, NSTL, MS 39529-5004) and Joseph E. Murphy (Departmentof Physics,Universityof New Orleans,New Orleans, LA 70148) 2:14 M12. Laboratoryscaleinvestigationof the amplitudeandphaseof signals reflected from sea ice ridges. T. McLanahan (Naval Air Systems Command,Washington,DC 20361-5220), O. I. Diachok, and S. C. Wales (Naval ResearchLaboratory,Washington,DC 20375) Acousticreflectivityfrom plasticmodelsof seaice ridgeson a thin platewereinvestigated for 0.67< ka < 6 (wherek is thewavenumber and a isthemeanridgedepth).Modelswereconstructed of lucite,whichhas acoustic velocities consistent with analyticallyestimated velocities of sea iceridges,andof"fast cast,"a materialwith muchlowervelocities. For ka > I measurements from randomlyorientedandparallelluciteandfast castridgesona thinplateandonan infinitehalf-space werenearlyidentical,andin goodagreement withTwerskytheoryprediction. At thelowest frequency, ka ---0.67,andatgrazingangles lessthan20ø,measured reflection losses weresignificantly greaterthanpredictions: twiceaslargefor randomlyorientedluciteridgeson a thin plate.The phasechangeupon reflection wasstrongly dependent onthematerialproperties of theridges and their orientation.Resultsfor randomlyorientedluciteridgeswere foundto be reasonably consistent with previously reportedinferences of reflection lossandphasebasedonmatchedfieldprocessing ofdatarecorded on a largeapertureverticalarray in the Arctic [Livingstonand Dia- Many factorsmake modelingunder-iceacousticpropagationand scatteringdifficult:Interfacesare roughwith randomprotusions; ice is acousticallypenetrable;little data exist;few validatedmodelsexist;frequencies of interestrangefromthehertzto thekilohertzregion.Thusno one modelappearssufficientfor all conditions.With this in mind, two under-iceacousticmodelsat oppositeextremesarc examined.One model is simplistic,usingKirchhoff approximation,ice reflectioncoefficients [S. A. Chin-Bing,J. Acoust.Soc.Am. Suppl.1 78, S57(1985)] andrms roughness. Model resultscomparefavorablywith high-frequency measurementsfrom a flat cylindricaliceblock.Implicationsare that simplistic modelsmaybeusefulin generatingsyntheticunder-iceacousticdatato augmentsparseexperimental databases.The othermodelissophisticated in thatit isa full-waverange-dependent acoustic propagation modelbased on the finiteelementmethod[J. E. Murphy and $. A. Chin-Bing,J. Acoust.Sac.Am. Suppl. ! 81, S9 (1987)] and appliedto an under-ice surfacesimulatedbyfractalssuperimposed onan icekeelrepresentation. Numericalresultsare presented for this low-frequency model. [Work supportedby ONR/NORDA. ] 2:30 chok,J. Acoust.Soc.Am. Suppl. 1 81, $85 (1987) ]. M 16.Low-frequency propagation acrosstheEastGreenlandandFrontal Zone:Depthdependence of transmission loss.LeonardE. Mcllberg,D. N. Connors,D. G. Browning,G. Botseas (Naval UnderwaterSystems 2:18 Center,Newport,RI 02841), and O. M. Johannessen (Geophysical Institute,Universityof Bergen,Bergen,Norway N-S014) MI3. Low-frequency transmission lossmodelingin the Arctic using measuredmode attenuation coefficients.T. C. Yang (Naval Research Laboratory,Washington,DC 20375} Low-frequency transmission lossin theArcticcanbeeffectively described by thereflection/scattering lossat theice-waterinterface dueto theupwardrefractive sound-speed profile;thesoundfieldthat interacts withthebottomisquicklyattenuated astobeunimportant at longranges. Variousmodelsfor the reflection/scattering losshavebeenproposedin theliteratureandcanbeshownto agreewith singlehydrophone databy adjusting theiceroughness or otherparameters in themodel.However,a criticaltestof the modelpredictions hasto be conducted usingvertical In a previous paper[ L. E. Mellbergeta!.,J. Acoust.Soc.Am. Suppl.1 78, S71( 1985) ] thedirectionaldependence of acoustic modesacrossthe EastGreenlandFrontalZonewasexamined. However,fora givensource andreceiverdepth,transmission losscanbedueto a combination of these directionallydependent acousticmodes.By changingthe depthsof the sourceand receiver,the directionaldependence of the transmission loss maybechanged, in somecasesto theextentof causinga reversalin the dependency. Resultsaregivenfor a seriesof sourceandreceiverdepths locatedat variousrangesrelativeto thefrontalzone.[Work supported by NUSC and ONR. ] arraydata,sincethedepthdependence of thesoundfieldprovidesa more meaningful measurement of the angulardependence of the reflection/ scattering loss.In thispaper,theangularandfrequency dependence ofthe reflection/scattering coefficients wasinvestigated, basedon themodeattenuationcoefficients previously deducedfromverticalarraydata.It was found that the measuredreflection/scattering losswas larger than most modelpredictions usingrealisticiceroughness parameters. A transmissionlossmodel,basedon a simpleparametricfit to the reflection/scattering data,wasshownto agreewell with previouslypublisheddata. S31 J.Acoust. Soc.Am.Suppl. 1.Vol.82.Fall1987 MI7. Relativescatteringfromdry cracks,iceridges,andleads.Jacques R. Chamuel(Sohoquest/Advanced Ultrasonics Research, P.O.Box153, WellesleyHills, MA 02181) andGary H. Brooke(DefenceResearch EstablishmentPacific, FMO, Victoria, British Columbia V0S lB0, Canada) 114thMeeting: Acoustical Society ofAmerica S31 Little is knownaboutthe characteristics of dry cracksin Arctic seaice and their effecton low-frequency underwateracousticwavesinteracting withtheice.Ultrasonicmodelingresultsarepresented showingthe relativescatteringfromdry cracks,leads,andridgesin floatingplatesboundinga longshallow-water waveguide. Backscattering fromsolidridgesand ridgeswith internalcracksrevealthepotentialimportance of dry cracks on low-frequency transmission lossprediction.In modelswith a large TUESDAY AFTERNOON, 17 NOVEMBER numberof rectangular ridges,greattransmission losses wereobserved at frequencies nearlycoinciding with twicethe ridgewidth.The articlefocuseson multiplescattering frominfinitelylongshallowparallelcracks, wherethecrackdepthismuchsmallerthanthewavelength. For a given numberof cracksor ridges,theattenuation of lowfrequencies propagatingin thewaveguide isgreatlyaffected by theirspatialdistribution along thewaveguide boundary.[Work supported by DREP andONR•] 1987 POINCIANA ROOM, 2:00 TO 4:15 P.M. SessionN. ArchitecturalAcousticsII: AcousticalDesignof RenovatedPerformanceFacilities II David Lubman, Chairman DavidLubmanand Associates, 14301MiddletownLane, Westminster, California92683 Chairman's lntroduetionw2:00 Invited Papers 2.'O5 N1. The imimetof Playhouse Squareon the resurgence of Cleveland.LawrenceWilker (PlayhouseSquare Foundation, 1501 Euclid Avenue, Suite 810, Cleveland,OH 44115) Cleveland,Ohio, wasa city in economicdistresswhena groupof farsightedcitizensand local officials decidedto developa downtownculturalcenter.Lackingthe fundsfor a 300 milliondollarproject,it was decidedto renovatea three-roomvaudeville/filmthefitrecomplex,knownasPlayhouse Square.Onetheatre wasdesignated asa drama/dance/chamber orchestra facility,anotherasthehomeof theCleveland Operaand BalletCompanies, andthe third asa venuefor popularentertainment. The papertracesthe historyof the projectandits impacton the resurgence of Cleveland. 2:30 N2. New theatresfor old. Norris Strawbridge(Borrelli FrankelBlitsteinSasakiAssociates, Inc., 7000 S.W. 62nd Avenue, Suite 520, Miami, FL 33143) TheMiamiBeachTheatrefor thePerforming Arts,recentlyrenamed TheJackieGleasonTheatre,began itslife asa multipurpose theatre/sports facility.In themid-1960's a combination of communityneedsanda decision byJackieGleason to broadcast histelevision programs fromMiamiBeachresulted in a renovation of the hall asa performingarts/televisionstudiovenue.Althoughthe theatrelackedcertainamenities,it fostered thegrowthofoneof thelargestooncert/opera/dance series in Americaandhasbecome thehomefora 26-week Broadway showseason. Recognizing theinherent problems in thefacility,theCity recently commissioned a second renovation tobringthespace uptocontemporary acoustic andtheatre standards. Thepaperdiscusses themethods usedto developan architectural programfor therenovation andthesolutions developed by the designteam to attaindesiredgoals. 2:55 N3. Renovation of the OrpheumTheatre,Vancouver,Canada.Ewart A. Wetherill (Wilson,lhrig and Associates,5776 Broadway,Oakland,CA 94618) Despite theever-increasing costofperformance facilities, thedesign andconstruction ofa completely new auditorium ispreferred withremarkable consistency tothealternativ.e ofrenovating or remodeling anexisting unused, orunder-used, auditorium.However,thereuseofanexistinghahmayyield--in additionto substantial costsavinga---such advantages asprimelocation,existingsocialrecognition, andhistoricalsignificance. This paperreviewstheremodeling of theOrpheum,originallya vaudeville theatre,fromcinematoconcerthallasan alternativeto its conversion to severalsmallcinemas.In mostrespects, the renovationhaspreserved the originalarchitectural characterwhileattaininga criticallyacclaimed performance facility. $32 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyo!America $32 3:20 N4. Acoustics of thenewFllcoeCenter,Wolf Trap FarmPark for thePerformingArts.ThomasR. Hotrail andRein Pirn (BBN Laboratories, Incorporated, 125CambridgePark Drive,Cambridge, MA 02140) To replace theoriginalFileneCenterthatwasdestroyed byfire,a newstructure wasdesigned andbuiltin a scant2 years( 1982-84). Like itspredecessor, the 3800-seatcoveredhouse,with an additional3000seatson the lawnbeyond,servesasa summerhomefor theNationalSymphonyOrchestraandcatersto manyotherforms of artisticproduction.Undera mandateto rebuildthestructureasit hadbeen,the acoustical optionswere limited.Knownproblems suchasdiscreteechoesoff the highceilingwereeliminated.Stageacoustics were greatlyimproved by incorporating a totallynewshell.Mostsignificantly, thecenterwasequipped withstateof-the-artsoundsystems, includinga sophisticated enhancement systemthat coversthe under-balcony areas. The principalacousticalcharacteristics of thecenterarediscussed andthe soundsystems, their purpose,and their operationalparametersdescribed. 3:45-4:15 Panel Discussion TUESDAY AFTERNOON, 17 NOVEMBER 1987 TUTTLE Session O. Structural Acoustics and Vibration NORTH ROOM, 2:00 TO 4:05 P.M. II: Fluid-Loaded Structures Albert J. Tucker, Chairman Officeof Naval Research, MechanicsDivision,Code1132SM,800 North QuincyStreet,/trlington,Virginia22217 Chairman's Introductlonm2:00 ContributedPapers 2:05 O1. Simulated wave-vector filtering of impact-generatedflexural wave disturbanceson a finite thin-walled cylinder. Allan D. Pierce, Jerry H. Ginsberg, and Gary P. Schwaiger (School of Mechanical Engineering, GeorgiaInstituteof Technology, Atlanta,GA 30332) The authorshaverecentlyundertakena theoreticalsimulationof ex- perimentssimilarto thoseperformedrecentlyat NRL by Earl G. Williamsand his colleagues, in whicha submerged cylinderis transiently excitedandnormaldisplacements areinferredusingthegeneralized nearfieldacoustical holography(GENAH) technique. In thepresentpaper, Previous workby theauthorwith PierceandGinsberg[J. Aeoust. Sac.Am. Suppl.1 79, S35(1986)] usinga variationalformulalion has beenextended to axisymmetric bodiesin a generalstateof axisymmetric vibration. A Rayleigh-Ritztechnique is usedto calculate theacoustic pressure distribution onthesurface of axisymmetrically vibrating finite cylinders. Thecomputed surface acoustic pressure andthegivennormal surfacevelocityarethentakenasinputsintoanintegralrelationfordeterminationof the farfieldradiationpatterns.Differenttrial functionsare the cylinderis modeledas a thin-walledshell with rigid immovableend employed in theexpansion fortheunknown surface pressure andresults arecompared witheachother.It isdemonstrated thatthecomputational solutions converge muchfastertowardthecorrect answers ifoneincorporateshisphysical insightandpriorknowledge intotheconstruction of capsandwiththeendconditions takenasbeingthatof simplysupported. A shortdurationpoint forcef(t) is appliednormallyto the surfaceand oneseeksto describe theresultingnormaldisplacement fieldusinga highfrequency(and shortwavelength)asymptotictheoryof propagating and evanescent flexuralwavepulseson a curvedplate.Fluid loadingis taken putedusingRogers' SHll'program.[Worksupported byONR.] intoaccountapproximatelywith modifications derivedusingthe assumption that the flexuralwavelengthsare short comparedto the lengthand radiusof the cylinder.One asksfor the time and surfacepositiondependenceof the normal surfacedisplacement within a local regionof the cylindricalsurfacebeforeand after a simulatedwavevectorfilteringhas beenapplied.The filteringprocessis suchthat one seeksto isolateonly that contributionfrom a narrowfrequencybandand from a localized regionof wavenumber space.The exercise isintendedto demonstrate that onecan isolateand followtravelingwaveletspropagating awayfrom the source,spiralingaroundthe cylinder,and undergoingsuccessive reflectionsat theendcaps.IWork supportedby ONR. ] 2:20 02. Numerical implementation of variational techniques for sound radiation from axially and radially vibrating finite cylinders. Xiao- Feng Wu (School of MechanicalEngineering,Georgia Institute of Technology,Atlanta, GA 30332) S33 J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 admissible trial functionsfor the unknownsurfacepressure. Examples studiedincludefinitecylindersvibrating,respectively, in theradialdirectionandaxialdirectionasa rigidbody.Numericalresultsagreewellwith thosepreviously published by Copicyandby FenIonandwiththosecom- 2:35 03. Impulseresponse of a continuously ribbedpanel.G. Maldanikand L Dickey(David TaylorNavalShipResearch andDevelopment Center, Bethesda. MD 20084) The impulseresponse functionof a uniformpanelimmersed in a uniform environmentis assumedknown.The modificationin this impulse response functiondueto theattachment of identicalandregularlyspaced ribsisderived.The mannerof negotiating thismodifiedimpulseresponse function from the discretenessto the continuum of ribs is discussed. The negotiationrevealsthe criteria that needto be satisfiedfor uniformly lurepingtheribsin thepanel.Whenthecriteriaaresubstantially satisfied, theribbedpanelbecomes essentially a uniformpanel.The ribsarethen accounted for by lumpedmodification of the uniformimpedance of the unribbedpanel. 114th Meeting:AcousticalSocietyof America S33 04. Vibration of a plate with distributed structural stiffness. Mauro Pierucci(DepartmentofAerospaceand Engineering Mechanics, SanDiegoStateUniversity,SanDiego,CA 92182) wavenumber and the wavenumber the scheme for finite boundaries and to formulate interface conditions with an acousticfluid. Qualitativeand quantitativeresultsare presented In mostanalysesof plates,only two typesof platevibrationshave receivedwideattention:homogeneous platesandplateswith discretediscontinuities.In thiswork,thestructuralinhomogeneity hasbeenselected to becontinuous anddistributed.The plateis assumedto beinfinite.The stiffness hasbeenassumedto vary in a sinusoidalmanner.The resulting differentialequationhasbeensalvedby theuseof Fouriertransformsand the solutionin the wavenumber spaceis presented. The plateadmittance is a function of the structural to theentirebody.The dispersion spectrumexhibitstheclassical Brillouin structureof stoppingand passingbands.Even thoughFloquet'stheory appliesessentially to infiniteregions,it is nevertheless possible to adapt for harmonicmotionsof two geometries, namely,a layeredhalf-space withan overlyingacoustic fluidanda submerged infinitelaminatedplate. Specialattentionis paid to the energypartitionbetweenthe layeredsolid and the fluid for the completerangeof frequencies. [Work supportedby ONR. ] of the distributedinhomogeneity. The resultsfor the in vacuoand the fluidloadedplatesarepresented. In theveryshortandtheverylonginhomogeneitywavelengthlimits,theeffectof thestructuralinhomogeneity is very weak at all structural wavenumbers. The main contribution to the struc- tural admittanceis presentwhenthe valueof thestructuralinhomogeneity wavenumberis closeto the coincidencestructuralwavenumber.For the in vacuoca•es,it is shownthat the effectof the inhomogeneityis limitedto valuesof thestructuralwavenumber equalto thecoincidence wavenumber plusor minusan integralmultipleof the inhomogeneity wavenumber. 3:05 05. Resonantresponseof elastically supported,fluid-loadedplates. JamelHammoudaand CourtheyB. Burroughs(Engineering Science and MechanicsDepartment,The PennsylvaniaState University, UniversityPark, PA 16802) An analyticmodeloftheresponse offiniteplatessimplysupported on two opposing edgesandelastically supported on theotheredgesis derived.Transverse shearandrotaryinertiain theplateandadjustable stiffheSS constantsin theelasticsupportsare includedin the model.The elastic 3:35 O?. Accelerometerarray ns a meansto evaluate structural anderw•ter sound.D. Brenot (Department of Physics,Oncra, BP ?2, 92322 Ch•tillon Cedex, France) The turbulentboundarylayer 1ow-wavenumber component,an importantsourceof underwaterstructureencitation,cannotbe precisely determined.Nevertheless, the resultingwall accelerations can be accuratelymeasured. Theyalreadycontainthecouplingwith thefluid,avoid thecomputationof the structuralmotion,andallowdirectcalculationof the farfieldnoiseradiatedby the vibrationitself:The singlefrequency farfieldnoiseradiatedin a givendirectionby a baffledpanelcomesfrom onecomponentof the wavenumberacceleration spectrum,and only one ( seeM. C.Junger andM. P6rulli,Elements d '•4coustique Physique, p. 59). Thiscomponentmaybeobtainedexperimentally byperformingtheFourier transformof the wall acceleration usingan accelerometer array.An underwater experiment isshownona 3-ram-thickpanelwitha 5 X 5 acceIcrcmeterarraymatrix (Ax = 70 min. Ay = 60 ram) andona thickcylinder (radius:i m; length:12 m) with a 10-accelerator lineararray.The paraboliclawsbetweenwavenumber and frequencyclearlyappearand are quite typicalof eachelement.Thesemeasurescan thus be usedto estimatetheir radiation efficiency.[Work supportedby DCN/CERDAN. i supportstiffness constantsare adjustedto simulatesimplysupported, clamped, andfreeboundary conditions, andtheresultsarecompared to published results.The variationof resonant frequencies with the stiffnesses of theelasticsupports isshown.Theeffectof fluidloadingonthe resonant responses of theelastically supported plateis presented. 3:50 08. New approachesin vibroaconsticmeasurement systemdesign. Joseph A. Clark andPaul M. Honke(David W. TaylorNavalShip Researchand DevelopmentCenter,Bethesda,MD 20816and Catholic University,Washington, DC 20064) 3:20 Recentadvances in measurement instrumentation andin computer 06. Dynamicresponse of submerged layeredcomposites. Arthur Braga andGeorgeHerrmann(Divisionof AppliedMechanics, Departmentof MechanicalEngineering,StanfordUniversity,Stanford,CA 94305) hardwaresuchas morefully programmable signalconditioners, faster data convertors, processors and otherhandlers,higherresolution displays,andlargermemories invitethedevelopment of newvibroacoustic measurement approaches. In thistalk,thedesign andimplementation ofa Thedynamicinteraction between elasticandhomogeneous two-phase periodicallylayeredbodiesand a surroundingacousticfluid has been studied.The plane-strain elastodynamic behaviorof the compositehas beenmodeledonthebasisofFloquet'stheoryforperiodicstructures. This approach consists ofsolvingtheequations of motionfora cellthatrepeats periodically,andthenapplyingFloquet'stheoremto extendthe solution newsystem for measuring radiatedandself-noise pathsassociated with complex,large-scale, immersedstructures will be described. New approaches willbepresented in thecontextofsoftware developed or utilized to achieveimprovedperformance of control,processing, display,and analysis functions. Advantages of thenewapproaches forstudies ofmore S34 J. Acoust. Soc.Am.Suppl.1, Vol.82,Fall1987 subtleand complexvibroacoustic effectswill be discussed. 114thMeeting: Acoustical Society ofAmerica S34 WEDNESDAYMORNING,18NOVEMBER1987 UM AUDITORIUM,8:10TO 11:45A.M. Session P. SpeechCommunication III: Recognition Jay G. Wilpon, Chairman /1T&TBellLaboratories, Murray,Hill, NewJersey 07974 Chairman's Introduction--8:10 ContributedPapers 8:15 PI. Performance benchmarks for isolatedwordrecognition. Lawrence R. RabinerandJayG. Wilpon(AT&T BellLaboratories, MurrayHill, NJ 07974) The performance of isolated wordspeech recognition systems has steadily improved overtimeasmoreislearned abouthowtorepresent the significant events in speech andhowtocapture these events viatraining algorithms. In particular, algorithms based onbothtemplate matching andhidden Markovmodels (HMMs) havebeendeveloped thatyieldhigh accuracy onseveral standard vocabularies including thetendigits(zero tonine)andthesetofletters oftheEnglish alphabet. Results arepresentedherethatshowtheeffectofanalysis parameters (e.g.,framesize,frame rate) on recognition performance for a speakertrainedsystemand the performance of both templateand HMM-basedrecognizers (both speaker trainedandspeaker independent) onthedigitsandthealphabet Techniques thatautomateall or partsof thisprocess arethereforehighly desirable.Two approaches havebeenused:the generationof pronunciation networksfrom baseforms by applicationof phonological rulesand theabstraction of suchnetworks fromtranscription data.Thispaperdescribes a semiautomatic procedurefor derivinglexicalmodelsfrom manual phonetictranscriptions. The algorithmusesa feature-based distance thatspecifyalternatepronunciations for words.The paperdescribes the results ofconstructing lexicalnetworks fora 1000-word vocabulary, using a corpusof 10000 work tokens.It shouldbe notedthat the technique metric to align differe pronu and prod pho ne describedhereis not restrictedto manualtranscriptions, but can be extendedto automatictranscription suchastheoneproducedby a front-end componentof a speechrecognitionsystem. vocabularies. 9:00 P4. A wordclassification methodfor therecognition of Japanese froma largevocabulary. KazunagaYoshida(C&C Information Technology Research Laboratories, NEC Corporation, 4-1-1,Miyazaki,Miyamae- 8:30 ku, Kawasaki213,Japan) P2. Classificationof fluent speechin coarseacousticclasses.PaulEric Stern (IntelligentSystemLaboratory,Xerox PoloAlto Research Center,3333 CoyoteHill Road, Palo Alto, CA 94304) This work is motivatedby resultsof recentstudieswhichshowthat lexicalaccess usingrobustbroadphoneme classes canpotentially resultin smallsetsof candidateworkdsfrom a largelexicon,and thusis effective Broadphoneticclasseshavebeendemonstratedto be usefulfor con- forlargevocabulary wordrecognition. In thispaper,a setof broadpho- straining search duringlexicalaccess butnecessitate thattherecognition ofbroadphonetic features beachieved inanerror-free mode.Thisgoalis nemeclassessuitablefor Japaneseand a new multilevellexicalaccess modelare proposed.First, broadphonemeclasses that haverobustfea- notrealisticduetotheabstract definition of phonetic classes. However,a similartaskmaybeperformed throughtherecognition ofacoustic classes that are subsequently mappedto phoneticcategories. A coarseacoustic recognizcr hasbeenimplemented in thecontextof a multispeaker fluent speech recognition system. It furnishes robustacoustic representations corresponding to differentdegrees of granularity of speech thatwill constrainfurtherprocessing onthesignal. Thisclassifier hasbeenimplementedusinga Bayesian technique allowingtheapplication ofdifferentkinds of classrecognition decisions. The parameters usedto represent the speech signalaredependent ontheacoustic classes andrangefromwhole spectrum characteristics to specific discriminating features. Thechoiceof acoustic classes isdiscussed, onemaincriterionbeingthemaintenance of turesforJapanese areinvestigated, andtheirconstraining powerisexaminedusinga dictionarycontaining over20 000 words.Next,a multilevel lexicalaccess modelisproposed basedonourstudyaboutthebroadphonemeclass.The modelconsists of fourstages: syllablenucleus detection, consonantclassification,syllable nucleus classification,and detailed phoneme classification. Thelexicalsearchspacecanbesignificantly reducedusingtheaccess model.Finally,a wordclassification methodbased ontheaccess modelisalsoproposed. Results ofrecognition experiments wereencouraging.[This studywasconductedwhile the authorwasvisit- ingat the Research Laboratory of Electronics, MIT, Cambridge, MA 02139.] a correspondence betweenacousticand phoneticlabels.Becauseof the variability of thesignal,butalsoof thepropensity to errorof anyrccognizer,multipleandcontextual mapping isadvocated. Results arereported and commentedupon. [Work supportedin part by INRIA and DARPA. ] P$. Appllcationof hiddenMarkovmodelsto automaticspeech endpoint detection.J. G. Wilponand L. R. Rabiner(AT&T Bell Laboratories, Office2C-573,Murray Hill, NJ 07974) 8:45 P3. Using featuresto empirically generatepronunciationnetworks. AlexanderI. Rudnieky(Departmentof ComputerScience,CarnegieMellon University,Pittsburgh,PA 15213) Theconstruction ofhigh-quality lexicalmodels forspeech recognition systems isa labor-intensive process, requiring notonlya greatdealoftime but alsoappreciableexpertiseon the part of the humanmodelbuilder. $35 9:15 J.Acoust. $oc.Am.Suppl. 1,Vol.82,Fall1987 Accuratedetection of theboundaries of a speech utterance duringa recordinginterval has beenshownto be crucial for reliableand robust automariespeechrecognition.The endpointdetectionproblemis fairly straightforwardfor high-levelspeechsignalsspokenin low-level station- ary noiseenvironments. However,theseidealconditions do not always exist.Oneexample,wherereliableworddetectionis difficult,is speech spokenin a mobileenvironment. Currently,mostendpointdetectionalgorithms useonlysignalenergyanddurationinformation to performthe endpoint detection task.In thispaper,anendpoint detection algorithmis 114thMeeting: Acoustical Society ofAmerica S35 detection schemeswere tested. The results showed that the HMM-based caltract.Thispropertyis in conflictwithlinearsource-filter models of speech production andan effortis madeto minimizetheeffectof the fundamental frequency F0 ontheanalysis result[e.g.,Hermansky etal., Proceedings IC.dSSP-83(IEEE, New York, 1983),pp. 778-780]. Our experiments showthatthehumanauditory perception hasproperties sim- approach to endpoint detection performed significantly betterthenthe energy-based system. Theoverallaccuracy ofthesystem usingtheHMM endpointdetector,whentestedon the 11 worddigitsvocabulary(zero throughnineandoh) with speech recorded in variousmobileenviron- ilar to the abovedescribed propertiesof automaticanalyses. The method of confusion scalingis appliedin pairedcomparison of syntheticvowellike stimuliwith varyingF0 andthesecondformantfrequency F2. The resultsshow that the perceptualestimateof formant positiongoes ments,was99.4%. A toprecognition performance of 99.7%,onthesame conditions,wasachievedby usingan HMM recognizerwith explicitendpointing. througha cycleof positive andnegative deviations astherelativeF2/FO frequencygoesfrom one integralnumberto the next. The positionis estimatedthe mostaccuratelywhenthe formantpeakfallseitherinto a harmonicpeakor if it liesrightbetweentwoharmonicpeaks.The maxi- presented thatisbasedon hiddenMarkovmodel(HMM) technology. Basedon a speaker-dependent speechdatabase from four talkers,and recordedin a mobileenvironmentunderfivedifferentdrivingconditions (includingwhiletravelingat 60 mphwiththefanon), severalendpoint mal deviation of the estimate due to harmonics of the F0 from the 120-Hz 9:30 P6. Sometechniquesfor erestingrobuststochasticmodelsfor speech recognition. Chin-Hui Lee (SpeechResearchDepartment,AT&T Bell Laboratories,600 MountainAvenue,Murray Hill, NS 07974) Recently,hiddenMarkovmodels(HMM) havebeenappliedsuccessfullyto bothisolatedandconnected wordrecognition. However,whenthe sameformulationis adoptedfor recognitionof moreconfusable vocabu- lariessuchasEnglishalphabets, therecognition performance isoftenless satisfactory. The mainreasonis that a moreaccuratemodelis required. Sucha modelshouldbemorerobustagainstsmalltrainingsamplesizes andshouldalsobeproperlyinitializedsothat thefinerfeaturesusedto discriminate confusable wordpairscanbeextracted.In thispaper,three specific robustness issues willbeinvestigated: choiceofobservation densities,modelinitialization,andincorporation of durationinformation.In a step-by-step attemptto address thoseissues, it wasfoundthat the same HMM formulationcan still be adoptedif acoustic/phonetic knowledge aboutthevocabularyistakenintoaccountin themodelparameterestimationandrecognition phases. Testingona 39-wordEnglishalpha-digitvocabulary,in a speaker trainedmode,indicates thattherecognition performancecanbesignificantly improvedandtheresultsarecomparable to the template-based DTW recognizer if modelparameters are properlyinitializedand durationalinformationis adquatelyincorporated. 9:45 neighborhood in the2-kHzsecond formantregionisaboutq-0.5%. Our findingsimplythat whenmodelingthespeech perception, asin theautomaticspeechrecognition, a certaindegreeof sensitivityof the analysis technique to F0 mightbebeneficial. 10:15 P9. The importance of contexton therealizationof phoneroes. Francine R. Chen (Xerox PARC, 3333CoyoteHill Road,Palo Alto, CA 94303) Linguists andspeaeh researchers haveformulated manyrulesdescribing the variabilityin the phoneticrealizationof spokenwords.Usinga databaseofspoken sentences thathasbeenphonetically transcribed (Lareelet al., "Speechdatabase development: Designand analysisof the acoustic-phonetic corpus,"DARPA SpeechRecognitionWorkshop, 1986),someoftheserulesarequantified byexamining, theeffectof differentphonereitenvironments on thephoneticrealizationof spokenwords relativeto a canonicaldictionaryrepresentation. Modificationsare analyzedin termsofthegeneralcategories ofinsertion, deletion,substitution, or match,andin detailin termsof eachtypeof modification. The different contextsare studiedalongmanydimensions, includingstress,foot position,partof syllable,andword-initial/word-final position.Representationsfor compactly describing theenvironmental contextsarealsodeveloped.Preliminaryresultssuggest that the importance of eachof the dimensions that canbe usedto specifya phoneme's contextis phonemedependent.Additionally,the strengthof the relationshipbetweenobservedmodifications andcontextsvaryacrossdimensions for eachphoneme.[Work supported in partby DARPA.] P7. Probabilisticestimationof lexical stress.GhassanJ. Freij (Speech Laboratory, Cambridge University Engineering Department, Trumpington Street,Cambridge CB2 1PZ,UnitedKingdom) 10:30 A probabilistic technique isdescribed for theautomaticestimation of lexicalstresspatternof isolatedwordsusinghiddenMarkov models (HMMs) with continuousasymmetricprobabilitydensityfunctions. Stress information ispotentially usefulforlarge-vocabulary speech recognition in that it facilitatesthe extractionof more robustphoneticdata withinstressed syllablesandleadsto moreefficientlexicalaccess. Adoptinga binarystressed-unstressed strategy, a constrained HMM network wastrainedusingobservation vectors consisting oftenacoustic measurementsofintensity, pitch,andspectral balance. Thesemeasurements were extractedfrom a corpusof disyllabicstress-minimal word pairs,each wordembedded in a phraseandrepeatedeighttimes.Further,theunderlyingtemporalstructureofthedataisrepresented implicitlybytheprobabilityofremaining in a statein relationtotheprobability ofexitingfromit. PI0. An integratedapproachto automaticdetectionof retroflexedand high-frontedapproximantsin continuousspeech.Roy Gcngcl and JamesHieronymus(Institute for ComputerSciences and Technology, Bldg. 225, A216, National Bureauof Standards,Gaithersburg,MD Using this technique, correctnessof classificationwas measuredto be ence."For example,a basicframe of referenceis the energyin the frequen- 89% for stressed syllablesand 81% for unstressed syllables. cy bandbetweenapproximately 2-3 kHz. Accordingto formanttheory, theenergylevelin thisbandshouldbe relativelylow for retroflexed approximants(F I-F3 below2 kHz) andrelativelyhighfor high-fronted approximants (F2,F3 above2 kHz for) malespeakers. In contrast,the energybandbetweenabout1.2-2 kHz shouldberelativelyhighfor retroflexedapproximants whilethebandbetween about0.6-1.6 kHz shouldbe 10:00 Pg. Should ASR front-end be insensitiveto fundamental frequency? (perceptualshift of formantpositiondue to fine harmonicstructureof voicedspeech).Hynek Hermansky(SpeechTechnologyLaboratory, 3888 State Street, Santa Barbara, CA 03105) Due to the fineharmonicstructureof the sourcespectrumof voiced speech, mostautomatic speech analysis techniques estimate formantpositionsthat systematically deviatefrom the resonantfrequencies of the vo- S36 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 20899) Our goalis the development of speaker-independent, semivowel detectorclassifiers for usewith continuous speech[ R. W. GengelandJ. L. Hicronymus, J. Acoust.Soc.Am. Suppl.1 80, S19(1986)]. Progress on detectionof retroflexedapproximants (/r/,/x/,/*ff) in onebranchof a classification tree, and high-frontedapproximants(/y/,/i/) and highfrontvowels(/I/,/el/,/•/) in another,isreported.The analyticalscheme reliesheavilyon energyratiosselectedon the basisof "framesof refer- low for high-frontedapproximants.By using"energy2000-3000" aspart of thedenominator for a retroflexiondetectorandasa partof thenumera- tor for a high-frontingdetector,energyratiosare developed. The quotients of theseratios are then usedin two-dimensionalcluster analyses. Otherratiosareusedin subsequent clusteranalyses. Resultssofar indicatethatgreaterthan90% of targettokenscanbecorrectlydetected in a corpusof severalhundredreadsentences. Ratiosof correctdetections/ 114th Meeting: Acoustical Society ofAmerica S36 incorrectdetectionsvary betweenapproximatelyi.0/0.63-1.0/2.50. Someprosandconsof themethodwill bepresented. [Work supported in part by DARPA ContractN0003984PD41304.] 11:15 PI3. Speaker-independent automatic vowelrecognition based onoverall spectralshapeversusformants.StephenA. Zahorianand Amir J. Jagharghi (Department of Electrical andComputer Engineering, Old DominionUniversity,Norfolk, VA 23529) 10:45 PII. Across-speaker and within-speakervariability of British English sibilant spectralcharacteristics.P. F.D. Seitz, R. A.W. Bladon, and L M. C. Watson (Phonetics Laboratory, University of Oxford, 41 WellingtonSquare,OxfordOXI 2JF, England) Testinghypotheses regardingthe potentialasphonemediscriminators of certainpropertiesof"physical" and "auditorilytransformed"sibilant spectra[A. Bladonand F. Seitz,J. Acoust.Soc.Am. Suppl.I 80, S19 (1986)], thepresentstudydrawsona database of 20 maleand20 female talkers,containing over2000tokensof/s/,/j'/,/z/, and/3/in a variety of phoneticcontextsandspeakingstyles.Two automaticparameterextraction routines were used to obtain measurements,for each token, of physicalandauditorilyscaledvaluesof themainspectralpeak,gradients of the leadingand trailing spectraledges,and bandwidthof the peak. Especially in females'productions of thespicsis,largeandcharacteristic variationsin theseparameterswereobservedas correlatesof individual talker,phoneticcontext,andstyle.Overviewsof theenvelopes of physical and auditoryvariationare given;baseduponthis knowledge,a speakeradaptivestrategyfor discrimination of BritishEnglishsibilantsis proposed.[Work supported by theAlveyProgramme.] Automaticrecognition experiments wereperformed tocompare overall spectral shapeversusformants asspeaker-independent acoustic parametersfor vowelidentity.Stimuliconsisted of four repetitionsof 11 vowels spoken by 17 female speakersand 12 male speakers (29, I 1,4 = ! 276totalstimuli). Formantswerecompul•.dautomatically bypeakpickingof 12th-order LP modelspectra. Spectral shapewasrep- resented usingthreemethods: ( 1) bya cosine basisvectorexpansion of thepower spectrum: (2) astheoutputofa 16-channel, l/3-octfilterbank; and(3) astheoutputofa 16-channel reel-spaced filterbank.Automatic •cognitionwasbased onmaximum likelihood estimation in a multidimensional space.For all casesconsidered, therepresentations basedon spectal shape resulted in significantly higherrecognition accuracy than for recognition basedon onlythreeformants. For example, usingthe entiredatabase of all speakers and 11vowels,recognition basedon spectral shapewasabout85% vs69% for threeformants. If thedatawere restricted tofemale speakers andtheseven vowels/a,i,u,•,a.,1,•/, recognition wasabout97% basedon spectralshapeversus84% for formants. Theseresultsindicatethat, at leastfor automaticrecognitionof vowels, spectral peakdetection is neithernecessary norsufficient. [Worksupportedby NSF.] 11:00 P12. Automariedetectionfor diphthongs.Hisao M. Chang (NYNEX Corporation,70 WestRedOak Lane,WhitePlains,NY 10604) It is generallyrecognizedthat diphthongsdo not have steadystates andonly displayidentitiesthrougha dynamicshiftof their spectra.In Miller's auditory-perceptual theory,dynamicshiftsof diphthongspectra arerepresented aspathson a "vowelmap" [1. D. Miller, J. Acoust.Soc. Am. Suppl.1 81, S16 (1987)]. From theutterances of Englishdigitsby two maleand two femalenativespeakers,an auditory-perceptual transformationsystemand a segmentation rule wereusedto transformthe spectralprominences contoursderivedfromthoseutterances containing diphthongs/al/and/el/inpathsonthevowelmap,andit wasfoundthat all pathsfrom thesamediphthong,independent on talker'ssexand phoneticcontexts,travelin the samerouteand form a "diphthonggig." After the detailedanalysisof the paths,an automaticdetectionalgorithmwas developed basedon angles,durations,lengths,origins,and otherattributesof thepaths.Whenapplyingthealgorithmto digitutterances from 21 newtalkers,therecognition accuracies for/al/and/eI/were 75% and 90% while the insertionrates were only 1.1% and 4.5%, respectively. [Work supported byNINCDS anddoneat CentralInstitutefortheDeaf, St. Louis MO.] S37 d. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 11:30 P14. Formant tracking using statistical linttern recognition.Robert T. GayvertandJamesHillenbrand(RochesterInstituteof Technology ResearchCorporationand RochesterInstitute of Technology,75 HighpowerRoad,Rochester,NY 14623) Formanttrackersthatusepeak-picking algorithms andad hoerules tend to sufferfrom occasionalseveremistakesin labelingformants.Re- cently,Kopec[IEEE Trans.Acoust.Speech SignalProcess. ASSP-34, 709-729 (1986)] useda statisticalapproachbasedon hiddenMarkov modelsto avoidthesekindsof problems. Thisstudyusesanotherstatisticaltechnique basedondiscriminant analysis. A modelistrainedonhandmarkedformanttrajectories and variousfeaturesfrom a collectionof utterances. The same features are then extracted from a test utterance and usedto determinethemostprobableformantfrequencies. Any combinationof features,includingpeaksfrom LPC andDFT spectra,canbeused withthistechnique. Preliminaryresultsindicatethat thismethodavoids seriousmistakes.Resultsof evaluatingthe formanttrackeron a large numberof trainingandtestsetswill be presented.[Work supported by RomeAir Development CenterundercontractF3060285-C-0008. ] 114thMeeting:Acoustical Societyof America S37 WEDNESDAY MORNING, 18 NOVEMBER 1987 BOUGAINVILLEA ROOM, 8:30TO 11:35A.M. SessionQ. EngineeringAcousticsII: SpecialSessionon DirectionalMicrophoneSystemsfor Airborne Sound JamesE. West, Chairman AT&T Bell Laboratories,600 Mountain Avenue,Murray Hill, New Jersey07974-2070 Chairman's Introduction---8:30 Invited Papers 8:35 QI. A unitarymultielement ultrasonic arrayforfocused imaging in air. LeslieKay (220Boylston Street,Unit 9007, Boston,MA 02116) A condenser microphone isdescribed thatincorporates 20 independent conductive elements coupledto a single laminardiaphragm. Theelements areseparately excited byanelectrical signal, delayed intimerelative to adjacent elements, to enabletheformationof a focused beamwithintheclosenearfield of thearray.Usinga radiationfrequency coveting anoctavebandwidth from100-200kHz, a spatialresolution of0.2in. parallelto theapertureand0.1 in. perpendicular to theapperture at thepointof maximumfocushasbeenachieved. No apertureshadinghasbeenincorporated at thisstagesothat thediffractionfieldcanbeexplored.Sincethisis formedbya verywidebandwidthradiation,anunusualfieldplotwasexpected. The resultsarediscussed anda videoshowsexamplesof the highresolutionrealtime imagingin air that cannowbeachieved.This hasmany applications, particularly in automation asanalternative toopticalvisionwhendistance toanobjectisessential information. 9:00 Q2. A review of adaptive microphonearrays. M. M. Sondhi (Room 2D-536, AT&T Bell Laboratories, Murray Hill, NJ 07974) In this talk, severalalgorithmsfor adaptiveoptimizationof broadbandmicrophonearrayswill be described.The arraysmay be linear,two-dimensional, or three-dimensional. Eachmicrophonein the array feeds a tappeddelayline filter,and theoutputsof the filtersaresummedto givethe finaloutput.The problemto be discussed is that of adaptivelyadjustingthe tap weightsof the filtersin order to optimallyreceivea desired speech signalin thepresence of broadband noisesources withunknownlocations andpossibly slowlyvarying characteristics. 9:25 Q3. Directionalprocessing for interferencereductionin hearingaids.W. M. Rabinowitzand P.M. Peterson (ResearchLaboratoryof Electronics,MIT, Room 36-761,Cambridge,MA 02139) Listenerswith hearingimpairmentsoftenhavegreatdifficultyin interpretingacousticenvironments having multiple sourcesand/or reverberation.Directional processing,obtainedby weightingthe outputsof multiple sensors,offerspotentialfor reducingsuchinterference,althoughcosmeticconsiderations imposefundamentallimitationson performance.Infixed processing, weightsare held constantto achievea fixedantenna pattern(pointingstraightahead)andarechosenasa compromise betweenclassical andsupergain limitssoas to maximizedirectivitywith tolerablesensitivityto receivernoise.A particularsecond-order gradientmicrophonesystem,with a sensorspanof 4.1 cm and formedsimplyby subtractingthe outputsof two cardioid capsulehearing-aidmicrophones, exhibitsa directivityof 8.5 dB, whichis nearoptimum(9.0 dB) for its noise sensitivity.In adaptiveprocessing, weightsare continuallyadjustedaccordingto propertiesof the environment. An adaptivealgorithm usingtwo omnidirectionalmicrophonesseparatedby 20 cm ( • headwidth) has beenevaluatedwith a targettalkerstraightaheadandan interference sourceat 45øin threecomputer-simulated environmentswith zero, moderate,and severereverberation.Both analyticmeasuresand intelligibilitytests indicatesubstantialreductionsin interferencein zeroand moderatereverberationand no reduction,or degradation,in severereverberation. Additionaland morerealisticevaluations of thesefixedand adaptivesystems arerequiredto determinetheirsuitabilityfor practicalhearingaids.[Work supported by NIH. ] 9:50 Q4. Automatic speech-seeking microphonearrays for teleconferencing.G. W. Elko (Room 2C-574, AT&T Bell Laboratories,Murray Hill, NJ 07974) The effectsof reverberationon speechtransductionin largeroomscanseverelylimit the intelligibilityof the desiredspeedsignal.In isotropicacousticfields,asidefrom locatingthe microphonein closeproximityto the S38 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 114thMeeting: Acoustical Society ofAmerica $38 speech source,it is necessary to develophighlydirectionalmicrophone systems. Concomitant with microphonearraydesigns for teleconferencing is theneedto steerthearrayat thedesiredspeech sourcein orderto maximizethe direct-to-reverberant signalenergy.The problemthat will be discussed is that of designing broadband (speech bandwidth)microphone arraysthatarecapable ofdetecting a speech source andautomaticallysteeringthe array to the directionof the source. Cont•butedPapers 10:55 10:15 Q5. Three-dimensional microphone arrays.J. L. Flanagan( Information PrinciplesResearchLaboratory,AT&T Bell Laboratories,Murray Hill, Q7. Characterization of the envelopeprotecting an acousticdetection system. D. Vaucher De La Croix, M. Lathuili•re, and L. P6rier NJ 07974-2070) ( MetravibR.D.S., 64 ChemindesMouilles,69130Ecully,France) Low-cost, high-qualitymicrophonesand integratedelectronics promptexaminationof sophisticated arraysfor soundpick-upin large enclosures. A three-dimensional, uniform configurationof receiversis studiedandshownto haveseveralbenefits.Beamforming anddelaysteeringcanbeaccomplished overthe4•rsolidanglewithoutspatialambiguity. Beamwidthisessentially independent of steeringdirection.As in ar- In a previouspaper,a schemefor computingthe radiationperturbation dueto a two-dimensional envelopeof arbitraryshapeenclosing an raysof fewerdimensions, usablebandwidth is governed at the upper frequency limit by receiverspacing andat thelowerfrequency limit by lineardimensions. Harmonicnestingin threedimensions canextendthe bandwidth of uniquespatialdiscrimination to anydesiredrange.Range selectivity, additionalto uniquedirectionaldiscrimination, canbe realizedby delaysteeringto nonplanarwavefronts. acousticalsourcewaspresented[D. Fiehot and J. M. Parot, D.3-5, 12th ICA, Toronto (1986) ]. The formulation associatesstructural finite ele- mentsand acousticintegralequations. Thoughoriginallyaimedto solve fluid-structureinteractionproblemswith strongcouplingbetweentransducerand envelopefor ka valuesup to 30 (k = acousticwavenumber, a = half-largestdimensionof the envelope),the codenamedABER2D wasappliedtodetermine theinfluence oftheenvelope shapeand/ormaterial on the directivityof the enclosedtransducerwith air asambientfluid. Deviationsfrom thefreefieldreference configuration havebeencomputedat low ka values(typicallyka _• 1), startingfrom a thin circularenvelope and progressively flatteningit to ellipticalshape.Suchcalculations haveorientedchoicesconcerninggeometryand certainmaterialcharacteristicsand resultedin designinga flattenedpolypropylene protection envelopeadaptedto an acousticdetectionsystem,whichdisplaysgood transparencyin the frequencyband of interest.[Work supportedby SEFT. ] 11:15 10:35 QS. Unidirectional,second-order•radient microphone.J. E. West, G. Q6. The evaluationof sourcestrengthusinga microphonearray over a M. Sessler, • andR. A. Kubli (AT&T Bell Laboratories, Murray Hill, hard reflectingsurface.Stewart A. L. Glegg and Richard C. Smart (Department of Ocean Engineering,Florida Atlantic University, Boca NJ 07974) Raton, FL 33431 ) This studyhasbeenundertakento measurethe noisesourceheighton highwayvehicles.A verticalseven-element microphonearray is usedand the methodof analysisis basedon the sourcebreakdowntechniquedevel- opedby FisherandTester.Thisiscapableof givingnarrowbandor «-oct sourcespectrafor eachindividualsourcein a distribution,with resolution down to zero frequency,providingthe approximatesourcepositionsare known, and there is a high signal-to-noise ratio. This methodhas been evaluated both in an anechoic chamber and in an outdoor environment over a hard surface. Good results have been obtained for sources which are0.13m apartdownto 125Hz. However,if thesourcepositions arenot exactly specified,the lower limiting frequencyis increasedto 500 Hz. [ Work is sponsored by theFloridaDepartmentof Transportation. ] S39 J. Acoust. Sec. Am. Suppl. 1, Vol. 82, Fall 1987 A second-order gradientmicrophone withunidirectional characteristicsis described.The microphoneconsistsof two commerciallyavailable, first-ordergradientelectret microphonesthat are properly baffledand arrangedwith their rotationalaxesin line. The outputof one of these transducers issubtracted fromthedelayedoutputof theothertransducer. The unidirectionalmicrophoneshowsa directionalcharacteristicthat is relativelyfrequencyindependent,hasa 3-dB beamwidthof the mainlobe of _+40 degrees,and exhibitssidelobes10-20 dB below the mainlobe. After equalization,thefrequencyresponse of themicrophone in itsdirection of maximumsensitivityis within :k 3 dB between0.3 and4 kHz. The sensitivity of themicrophone, includingitsequalization, is - 60 dBV/Pa at 1 kHz and the equivalentnoiselevel for the abovefrequencyrange amountsto 28 dB linear or 20 dB(A). • Permanentaddress:Technical Universityof Darmstadt,6100 Darmstadt,WestGermany. 114th Meeting: AcousticalSociety of America S39 WEDNESDAY MORNING, 18 NOVEMBER 1987 UNIVERSITY LECTURE HALL, 8:30 TO 11:30 A.M. SessionR. Psychologicaland PhysiologicalAcousticsIII: Temporal Perceptionand Comodulation Robert H. Gilkcy, Chairman ResearchDepartment,CentralInstitutefor theDeaf, 818 S. Euclid, Saint Louis,Missouri63110 ContributedPapers 8:30 9:00 R1. Discriminationof time-expandedtonal sequences.R. D. Sorkin and R3. Temporalaspectsof the auditory system:Gap detectionand the M.P. Snow(Departmentof Psychological Sciences, PurdueUniversity, West Lafayette,IN 47906) modulation transfer function. T. G. Forest and David M. Green (Psyehoacoustics Laboratory,Departmentof Psychology, Universityof Florida, Gainesville, FL 32611 ) Subjects werepresented with two sequences of tones,eachconsisting of eight50-mssinusoidal tones.Subjects had to discriminatewhetherthe sequences werethesameor differentbasedsolelyon the frequencypattern definedby eachsequence of tones.The correlationbetweenthe temporal envelopes ofthetwosequences, thevariabilityofsequence timing,andthe lengthof the intersequence intervalweremanipulated.In the "correlated"condition,thegapsbetweenthetonesin thesequences wererandomly variedacrosstrials(sequences withina trial hadidenticaltemporalenvelopes);in the"uncorrelated" condition,theenvelopes variedbothacross and within trials.The resultsreplicatedan earlierexperiment[R. D. Sorkin,J. Acoust.Soc.Am. (in press)]. Performance in the correlated conditionwasindependent of temporalvariabilitybut decreased with increasesin temporalvariability.This patternof resultsis consistent with theassumed dominance of "trace"and"context"modeprocessing in the twoconditions. In anadditionalmanipulation, all toneandgapdurations in thesecondsequence of eachtrial wereexpandedby 40%. Performance in the time-expanded conditionresembledthat foundin the correlated condition.This resultis interpretedas evidencefor a durationscaling process in auditorymemory.[Work supportedby AFOSR. ] A three-stage modelconsisting of a bandpass filterfollowedby a nonlinearelementanda low-pass filteris oftenusedto modeltemporalprocessing in theauditorysystem[e.g.,N. F. Viemeister,J. Acoust.Soc.Am. 66, 1364-1380 (1979)]. Two experimentalparadigms,the detectionof gapsin noiseand of sinusoidalmodulationof noise,have beenusedto estimatethe parametersof this model. A simulationof this model was usedto "listen" to waveformsin gapdetectionand modulationdetection experiments. Responses of thesimulatorwerebasedonstatisticalcalcula- tionsat theoutputof thelow-pass filterstageof themodel.Our decision statistic was the maximum-to-minimum ratio of the waveform from the low-passfilter.Goodagreement betweenthe dataproducedby normalhearinghumanobservers andthe simulatoroccurredwhenthe bandwidth oftheinitialfilterwas4000Hz andthetimeconstant ofthelow-pass stage was3 ms. [ Researchsupportedby NSF. ] 9:15 R4. Illusory continuity of periodic sounds:A eomparisanof tonal and infratonal temporal induction. Richard M. Warren, John M. Wrightson, and Julian Puretz (Department of Psychology, University of Wisconsin--Milwaukee, P.O. Box 413, Milwaukee, WI 53201 ) 8:45 R2. Proportional target-tone duration as a factor in the diseriminability oftonalpatterns.CharlesS. WatsonandGary R. Kidd (Departmentof Speechand Hearing Sciences,Indiana University, Bloomington,IN 47405 ) A seriesof experimentson listeners'abilitiesto extractinformation frompatternswithvaryingtotaldrrationsandnumbersof tonalcomponentshaspreviouslybeenreported[J. Acoust.Soc.Am. Suppl.173, S44 (1983); 77, SI (1985)]. In thoseexperimentslistenerswere testedin high-stimulus-uncertainty, same-differentpattern discriminationtasks, in whichthe tonalpatternsto bediscriminated differedby changesin the frequencyof one or more components. Discriminationperformancein Tonesmay appearto remain on continuouslywhen alternatedwith soundshavingappropriatespectralcompositionand intensity.The limiting conditionsfor this illusory continuity have been usedto definethe characteristics of neuralspectral(place) analysisof tones.The present studyextendsinvestigation oftheapparentcontinuityofinterruptedcomplexperiodicsoundsto infratonalfrequencies, exploringthe rangeof repetitionfrequenciesfrom 10-2000 Hz. Evidenceis presentedindicating thatreconstruction of obliteratedsegments of periodicsoundsneednotbe baseduponthe neuralplacemechanisms associated with perceptionof tones.It appearsthat illusorycontinuityof periodicitycaninvolvea perceptualsynthesis baseduponneuraltemporalinformation,and that ex- planations forthelimitsofillusorycontinuityoftones("pulsationthresholds") solelyin termsof a frequencydomainanalysisinvolvingneural placeinformationmay be incomplete.[Work supportedby AFOSR. ] thosetaskswasconsistentwith previousmeasuresof the frequencyresolv- ing powerof the auditorysystemwhenthepatternscontainedoneto three equal-durationcomponents,for total pattern durationsfrom 62.5-2000 ms. As the numberof componentswasraised,discriminationthresholds increasedby largeamounts,often by factorsof 10-100 for patternswith more than seven-eightcomponents.While this result might imply an informationallimit on pattern processing,it is also consistentwith the hypothesisthat targettonesare equallywell resolvedif theyoccupyequal proportionaldurationsof the patternsin which they occur. Resultsof a newexperiment,in whichproportionaldurationsof targettonesand the numberof tonesperpatternwereindependently varied,suggest thatproportionaldurationof the targettonesis in factthe primarydeterminantof patterndiscriminabilityfor tonalpatternsrangingfrom 100-1500ms in total duration.[Work supportedby AFOSR andNIH/NINCDS. ] S40 J. Acoust.Soc.Am. Suppl.1, Vol. 82, Fall 1987 9:30 RS. Critical bandwidth for modulation detection. William A. Yost (Parrely Hearing Institute, Loyola University, 6525 N. Sheridan, Chicago, IL 60201) Two sinusoidalcarriersseparatedby frequencyF were eachsinusoidally amplitudemodulated(AM). Subjectswereaskedto discriminate betweena complexin whichthe two AM signalshadthe samephaseanda complexin which the phasespectraof the lower frequencyAM componentsweredelayedrelativeto the upperfrequencyAM components. The 114thMeeting:AcousticalSocietyof America S40 thresholdvalueof thephaseshiftwasmeasured fromthree-pointpsycho- 10:15 metric functions obtained from a two-alternative, forced-choicetask. A broadband-noise backgroundand randomvariationin overalllevelwere usedto forcethe two normal-hearing listenersto attendto the temporal structureof the two-toneAM complexes. The frequencyseparationbe- R8. The effectof signal-frequencyuncertaintyin a CMR paradigm.John H. GroseandJosephW. Hall, III (DivisionOtolaryngology Head and NeckSurgery,Schoolof Medicine,Universityof NorthCarolina,Chapel tweenthe two carriers (F), the ratesof AM modulation,and the form of Hill, NC 27514) the phaseshift (entirewaveformor envelopedelay) wereparametersin the study.The resultsindicatethat the thresholdfor detectingthe phase The phenomenon of comodulationmaskingrelease(CMR) indicates that the auditorysystemis capableof comparingthe outputsof several auditorychannels simultaneously. Thepresence of thesignalispresumed to be cuedby a dissimilarityin the temporalpattern in that channel, relativeto the remainingcomodulatingchannels,due to the additionof the signal.In its simplestform, sucha cuingmechanismcouldoperate quiteindependently of anyapriori knowledgeof whichchannelthe signal occursin. To testthispossibility,a studywasundertakento comparethe effectof signal-frequency uncertaintyin a comodulatedanda noncomodulatedbackground.It washypothesized that the detectionof a signal whosefrequencyfluctuatedrandomly from trial to trial would be less deleteriouslyaffectedin a comodulatingbackgroundthan in a noncomodulatingbackground.A novel feature of this study was the use of amplitude-modulated pure-tonecomponents asthebackgroundmasker. Resultsto dateindicatethat the effectof signal-frequency uncertaintyis itself small relative to fixed-frequencysignals,and no significantdifferencehasemergedbetweenits measurementin comodulatedversusuncomodulatedbackgrounds. shift is small when the two carriers are within a critical band. However, listeners arestillquitesensitive to thephaseshiftsfor verylargefrequency separations betweenthecarriers.The resultswill bediscussed in termsof the auditorysystem's abilityto usetemporalmodulationto identifycommonsources in a complexsound.[Work supportedby AFOSR-Life Sciencesand the NINCDS.] 9:45 R6. Discriminationof envelopephasedisparity.GregoryH. Wakefield and Brent Edwards (Department of Electrical Engineeringand ComputerScience, Universityof Michigan,Ann Arbor,M148109-1109) Thesensitivity of theauditorysystemto changes in ongoingtemporal structuresacrossdifferentspectralregionsis investigated usingsinusoidally amplitude-modulated(SAM) carriers. Previous resultsare extendedfor thedetectionof phasedisparitybetweentheenvelopes at two carriers[J. Aeoust.Soc.Am. Suppl.181, S34 (1987) ] to thediscrimination of suchdisparitywhena nonzerorelativephaseoffsetbetweenthe envelopes is present.For a standardSAM two-carriercomplexwith a relativephaseoffsetof •b,the relativephasedisparityof the comparison •b+ A4 wasadjustedto measurediscrimination thresholdasa functionof •b.The resultsshowthatA•bvariedby morethana factorof 2 as•branges from0-180 deg.Modulationfrequency (8-64 Hz) or carrierseparation (500 Hz, 1-4 kHz) haslittle effecton this phasedependence. When ex- pressed asa change in envelope correlation (p), however, wefindthatAp is constantas a functionofp. Additionalexperiments usingharmonic complexes for envelopes supportthehypothesis that envelopecorrelation is the primary determinantof performance.A neural-basedcoincidencedetectormodelof envelopecorrelationis proposedto accountfor these findings.[Researchsupportedby AFOgR.] 10:00 R7. Evidencefor cross-channelprocessingin detectionof envelopephase disparity. Elizabeth A. Strickland, Neal F. Viemeister, Deborah A. Fantini (Department of Psychology,University of Minnesota, Minneapolis,MN 55455), and Margery A. Garrison (Departmentof CommunicationDisorders,Universityof Minnesota,Minneapolis,MN 10:30 R9. Comodulationmaskingreleasefor envelopedecorrelationcueversus envelopelevel differencecue. JosephW. Hall, III and John H. Grose (Division of OtolaryngologyHead and Neck Surgery, School of Medicine,Universityof North Carolina,ChapelHill, NC 27514) In pastinvestigations of comodulation maskingrelease(CMR) it was difficultto identifythe detectioncueusedby the listener,asseveraldifferent cueswereavailable.The presentstudyattemptedto isolatetwo cues: envelopecorrelationand envelopelevel.The signalwaseither a 10-Hzwidenoisebandcenteredon 500Hz, or a 500-Hz puretone.The maskers were 10-Hz-wide noise bands centered on 400 and 500 Hz (either co- modulatedor noncomodulated). In oneexperimentthesame10-Hz-wide noisebandcenteredon 500Hz wasusedasboththesignalandthe masker. Here, the signalresultedin no decorrelationof envelope,but did resultin anenvelope leveldifference. In theexperimentusingthe500-Hzpuretone as the signal,the signal resultedin a decorrelationcue; however,any envelopelevelcue wasrendereduselessin this experimentby randomly rovingthe levelof the flankingbandbetweeneachinterval,andby adjustingthenoiseleveloftheon-signal bandin thedummyintervalsto thelevel of the noise-plus-signal in the signalinterval.The majorresultwasthat substantial CMRs occurredboth whenthe envelopelevelcuewasavailable but the decorrelation cue was not, and when the decorrelation cue wasavailablebut the envelopelevelcuewasnot. 55455) The sensitivityto thecorrelationbetweentheenvelopes of twosinusoidally amplitude-modulatedsinusoldswas investigatedin two experiments. Correlation was manipulated by varying the phase disparity betweenthe envelopes. In eachexperiment,two carriers(either 1 and 2 kHz or I and 3.2 kHz) weremodulatedat the samefrequency(8 or 128 Hz) and addedtogethereither in phaseor, on signalintervals,with an adaptivelydeterminedphasedisparity.The startingphaseof theenvelope of the lowerfrequencycarrierwasrandomizedacrosstrials.In the first experiment,the lowerfrequencycarrierwaspresented at 65 dB $PL and the higherfrequencycarrierwaspresentedat levelsfrom 65 to 25 dB SPL. No decrementin thresholdwas noted until the higher frequencysignal was20-40 dB belowthe lowerfrequencysignal.In thesecondexperiment, the lowerfrequencycarrierwaspresentedat 65 dB SPL to oneear,andthe higherfrequencycarrierwaspresented at thesamelevelto the otherear. Thresholdswereslightlyabovethoseobtainedmonotically.Theseresults suggestthat subjectsare able to useinformationacrossfrequencychannelsto detectenvelopephasedisparityand do not needto useenvelope interactionsoccurringwithin a singlechannelin performingthis task. [Work supported by NINCD$ GrandsNS12125and NS07889.] $41 J. Acoust.$oc. Am. Suppl. 1, Vol. 82, Fall 1987 10:45 R10. Comodulation detection differenceswith multiple signal bands. BeverlyA. Wright andDennis McFadden(DepartmentofPsychology, Universityof Texas,Austin, TX 78712) Detectionthresholdswere determinedfor signalsconsistingof one, two, or fivenoisebandsembeddedin eight "cue" bands.The centerfrequenciesof the signalsrangedfrom 1250-3250Hz in 500-Hz steps,and thoseof the cuebandsrangedfrom 500-4000 Hz in 500-Hz steps.The temporalenvelopes of thecuebandswereeitherall correlatedor all uncorrelated.The envelopes of the multiplesignalbands( 1) werecorrelated amongthemselves as well as with the cue bands,(2) were themselves correlated,but not with the cuebands,or (3) were themselvesuncorrelat- ed. As in previous6omodulation experiments, thresholds weresimilarly poorwhenthe signalandcuebandswereeitherall correlatedor all uncorrelated.Detectabilitywastypically7-10 dB betterwhenthe signalbands were uncorrclated with the all-correlated cue bands. These results were no differentfor single-and multibandsignals.Thus the auditorysystemap114th Meeting:AcousticalSocietyof America $41 pearsbetterableto detectbothsimpleandcomplexsignalsin background havingthe same--ratherthan different--temporalenvelopes, aslongas the temporalstructureof the signal(s)is differentfrom that of the surroundingsounds.[Work supportedby NINCDS Grand NS15895.] ratherthanbynumber,andviceversa.Implications for theinterpretation of certainlanguage-learning disorders arediscussed. [Work supportedby NIH/NINCDS and AFOSR. ] 11:15 R12. Informationallimits in speechprocessingby normal and impaired listeners.Blas Espinoza-Varas, CharlesS. Watson,and Mary S. Kyle 11:00 (Speechand HearingSciences, Indiana University,Bloomington,IN R11. Rate andnumberaslimiting factorsin the perceptionof sequences of 47405 ) syllables.CharlesS. Watsonand Bias Espinoza-Varas(Speechand HearingSciences, IndianaUniversity,Bloomington,IN 47405) spokenlettersb, c,d, g,p, t, or v, produced asCVsendingin thevowel/i/. The durationof eachtokenwaseditedto 75 ms. Syllableonset-to-onset It hasbeenreportedthat someimpairedlistenerspreceivespeechas being"jumbled,or asgoingtoo fast,"whichsuggests that their informationalcapacitymaybe reduced.Recently,work with nonspeech stimuli [ WatsonandFoyle,J. Acoust.Soc.Am. 78, 375-380 (1985) ] hasshown that the amountof spectral-temporalinformationthat can be extracted from sequentialpatternscan be limited very stronglyby the numberof freelyvaryingcomponents includedin a pattern,and only weaklyby the durationof the individualcomponents. Usinga variationof the speech testdescribed in the previouspaper,informationcapacityof normaland impairedlisteners wascompared forsequences oftwo,four,orsevensyllables,presented at ratesof 6.5or 3.9 syllables/s. Percentinformationtransmission(IT%) was estimatedfor each temporalpositionof the sequences,number of syllables, and syllable rate. The sensorineural impairedlisteners hadnearlyuniform45-to-55-dBlosses, andagessimilar to thenormallisteners. Stimuliwerepresented at approximately constant times were 75, 150, and 300 ms. The listener's task was to transcribe the sensationlevels (30-40 dB) to minimize effectsof differencesin sensitiv- lettersin eachsequence, in a testformatin which180sequences were ity. The maximumnumberof syllablesyieldingIT% > 50% was,on the average,6-7 for normallistenersbut 4-5 for the impairedlisteners. [ Work supportedby NIH/NINCDS and AFOSR. ] Listeners'abilitiesto process the detailsof sequences of soundshave beenshowntobelimitedbytwomajorfactors:thenumberofcomponents in the sequence and the rate at whichthe components are presented.Indi- vidualdifferences in theabilityto process largenumbers ofcomponents at high presentationrateshavebeenproposedto accountfor certainlanguage-or learning-related disabilities. Little attentionhasbeenpaidto the possibleinteractionsbetweenthesetwo factors,nor to the distributionof theseabilitiesin the populationof normal-hearinglisteners.A test was developed tomeasure theeffects of thenumberof syllables and,independently,thoseof syllablerate,or listener'sabilitiesto process syllablesequences.The sequences consistof two, four,or sevennaturaltokensof the presented,varyingrandomlyin numberandrate.Resultsfor an initial test groupshowsomeindividual'sperformance to be morelimitedby rate WEDNESDAY MORNING, 18 NOVEMBER 1987 UMS 4 AND 5, 8:30 TO 11:45 A.M. Session S. UnderwaterAcoustics III: Three-Dimensional Oceanography andAcoustical Modeling Robert P. Porter, Chairman Department of ElectricalEngineering, FT-10, University of Washington, Seattle,Washington 98125 Chairman's IntroductionS:30 Invited Papers 8:35 SI. Three-dimensional oceanographic acousticmodelingof complexenvironments. W. A. Kuperman, MichaelB. Porter,andJohnS. Perkins(NavalResearch Laboratory, Washington, DC 20375) Remotesensing combinedwith oceanographic modelingopensup newpossibilities for quantiatively describingverylargeareacomplexoceanographic-acoustic phenomena. A complexoceanenvironment consists ofvariable oceanography including mesoscale phenomena andvariable topography suchascontinental slopes andseamounts. Predictingthe acoustical effectsfrom any to all of thesephenomena is a computationally intense problem whoseefficient andaccurate solutionwill,nodoubt,ultimately employnewandinnovative numerical physics. Asa stimulus todeveloping newapproaches, three-dimensional modeling results forsuch complexenvironments arebeinglookedat usinga technique thatcanrapidlycalculatepropagation lossovera widegeographical area.Thisapproximate technique, whichisbased onnormalmodetheory,hasbeenapplied to environments thatcontainbothvariableoceanography andtopography. Of primeinterestin theseinitial studiesis a searchfor graphicalpatternsthatemergefor complexenvironments wheretheacoustical results reflector "image"theoceanography andtopography. It isbelieved thatthesethree-dimensional representationsusingcolorgraphics to displayandconveycomplex ocean-acoustic phenomena will providetheimpetus to furtherunderstanding andcomputingtheinteractionof thesephenomena. $42 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S42 9:OO S2. Dynamicalforecastingof mesoscale frontsand eddiesfor acousticalapplications.Allan R. Robinson (HarvardUniversity, Divisionof AppliedSciences, Cambridge, MA 02138) Real-time nowcasts andforecasts ofoceanic mesoscale fields(velocity,pressure, temperature, soundspeed, etc.) arenowfeasible andhavebeeninitiatedinselected regions. Predicting the"internalweatherofthesea"is analogous to meterological weatherforecasting. Ourapproach isbasedonrecentprogress in phenomenologicalknowledge andmodeling capabilities andissystematic [A. R. Robinson, "Predicting openoceancurrents, fronts,andeddies,"in Three-Dimensional OceanModelsofMarineandEstuarineDynamics, editedby Nihoul and Jamart(Elsevier,Amsterdam,1987)]; i.e., it combines dynamicalmodel[A. R. Robinsonand L. J. Walstad,"The Harvardopenoceanmodel,"J. Appl. Numer.Math 3 (1987) ] andobservational estimates via data assimilation methods.The predictionproblemwill be overviewed,resultsfrom an ongoingforecast schemefor theGulf Stream(gulfcasting)will be presented, the couplingof dynamicalandacousticalmodels will bediscussed, andpreliminaryresultsfroma coupledsystemwill bepresented. 9:25 S3. High-resolutionoceanacoustictomogral)hy.Peter F. Worcester,Bruce Cornuelle,Bruce Howe, and Walter Munk (ScrippsInstitutionof Oceanography, Universityof Californiaat San Diego, La Jolla, CA 92093 ) The oceanacoustic tomography experiments conducted to datecanbedividedintomappingexperiments, in which many sourcesand/or receiversare usedto generatesufficientcrossingray paths to resolvethe mesoscale oceanstructure,andcueraging experiments, in whichdataareobtainedin a verticalslicebetweena singlepairof sources and/orreceivers in orderto construct horizontally averaged verticalprofiles. Whileit is economically feasibleto deployenoughinstruments to constructan arraycapableof mappingan areaa few hundredkilometers ona side,it becomes prohibitivefor areasonemegameter andlargerin scale,eventhough manyfewertomography instruments arerequiredthan conventional pointmeasurements. Onepossible ap- proachto makehigh-resolution, three-dimensional mapsof largeareasof theoceanis to develop a hybrid system,in whicha combination of mooredinstruments and preciselypositionedsourcesand/or receivers suspended fromshipsareusedtogenerate a dense setofcrossing raypaths.Simulations indicate thatfouror fivetransceivers mooredin a largearraytomakelong-term measurements oftheaverage oceanstructure could becombined withperiodiccruises withship-suspended sources and/orreceivers to giveexcellent mapsof a squaremegameter of ocean.It will beessential in suchexperiments to usedataassimilation techniques to combinethe dataobtainedasa functionof time by the shipinto a numericalmodelof the ocean.Other data types(e.g.,satellite altimeter data)canbeassimilated at thesametimeto improvetheresulting mapsof the ocean. 9:50 S4. The acousticstochastic inversepeoblem.Terry E. Ewart (AppliedPhysicsLaboratoryand Schoolof Oceanography, University of Wgshington, Seattle, WA 98105) It hasbeendecidedthatthestatistics of thephase,complexamplitude,andintensityfluctuations thatresult fromwaves propagating in mediawithrandomindexof refraction inhomogeneities canbepredicted. In fact, theemphasis hasbeenrefocused fromthedevelopment ofthetheorytoa requirement tounderstand morefully the statisticsof the oceanmedium.An excitingresultof the success of the theoreticalpredictionsis the possibility of invertingacoustic measurements to obtainthestatistics of the oceanmedia.In thestochastic inverseproblem,oneobtainscoefficients of statistical modelsof theacoustic indexor refractionusingoptimization techniques. Thesetechniques requirethat deterministic signals,e.g., tides,be treatedformallyin a differentmannerfromstochastic signals,e.g.,internalwaves.Examplesof theinverseproblemwill be given fromoceanandnumericalexperiments. Modelsoftheoceanicinternalwave,finestructure, andtidalprocesses, obtainedfrommeasured acoustic fluctuations, arecompared with thosemeasured. A discussion of thefuture expectations including3-D considerations will be presented. ContributedPapers 10:15 S5. Simulated reciprocal transmissionsthrough a three-dimensional sound-speed and current structure.JamesA. Mercer (Applied Physics Laboratory, College of Ocean and Fishery Sciences,University of Washington,Seattle,WA 98105) A computercodethat integratesHamilton'sequationsfor a continu- S43 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 ouslydefinedmediumwasusedto modela reciprocaltransmission experi- mentalonga 2000-kmpath.The description of themediumincludedthe three-dimensional sound-speed and currentstructuresof four mesoscale eddiesand a westernboundaryjet. Eigenraysfor the reciprocaltransmissionsare determined,displacements are compared,and the criticalphysical processes are identified.Travel time resultsare comparedwith those from lessrealistic two-dimensionaland one-dimensionalcodes. [Work sponsoredby ONT. ] 114thMeeting: Acoustical Society ofAmerica S43 10:30 $6. A discussion of three-dimensional wave propagation effects. Ding Lee, Paul D, Scully-Power,George Botseas(Naval Underwater SystemsCenter,New London,CT 06320), and William L. Siegmann ( RensselaerPolytechnicIntitute, Troy, NY 12180) In mostenvironments,the predictionof acousticwavepropagationis carriedout with two-dimensionalpropagationmodels.This shouldNOT misleadanyoneto believethat three-dimensionalmodelsare not needed. In thispaper,sound-speed informationfromeddymodelsisusedto examinethree-dimensional effects. This examination wascarriedout employing the IFD (two-dimensional)and the FOR3D (three-dimensional) models. A set of numerical results from these two models will be dis- cussed. Interestingfindingswill bereported.Thesefindingsprovidestimulation for continuedthree-dimensional modeldevelopment. specificsituation.In part ( 1) of this paper,the caseof individualmode propagation in SOFAR channelisinvestigated by analyzingthedifferent valid ranges.In part (2), the effectsof a mesoscale oceaneddyon mode propagation by usingPE (IFD) codearediscussed. The maingoalis to findout if the modecouplingmatrixis sensitive enoughto applyin the "matched-field"methodfor the purposeof oceantomography[Work supportedby NRC. ] 11:15 S9. Error estimatesfor 3-D computationalschemesof wavepropagation. GeorgeH. Knightly and Donald F. St. Mary (Center for Applied Mathematics and Mathematical Computation, Department of Mathematics,University of Massachusetts,Amherst, MA 01003 ) For severalcomputationalschemesin underwaterwavepropagation, 10:45 S7. Coherenceloss modelingusingthe PE method.F. D. Tappert ( Universityof Miami, Applied Marine Physics,Miami, FL 33149) Scatteringby volumefluctuationsof soundspeedand by boundary roughness causesa lossof coherenceof propagatingcw signalsin the ocean.A 3-D parabolicequationmodelfor the coherentGreen'sfunction (p) isderivedthatdescribes angle-and depth-dependent scattering loss. Byintegrating overa narrowsectorof azimuthalangles,it isshownthata 2-D verticalplanemodelgivesan excellentapproximation for forward scatteringloss.The resulting2-D ew modelis efficientlysolvednumericallyby generalizing thesplit-stepFourieralgorithmto performmultiple local Fourier transformsby windowingin depth, transformingfrom depthto verticalangle,applying anangle-dependent filter,summing over layers,andfinallyinverseFouriertransforming. For roughsurfacescattering,thismethodgivesa lossthat islinearin thegrazinganglefor small angles,in agreement with Marsh-Mellontheory. 11:00 $8. The PE methodand the oceantomography.E. C. Shang(NOAA, WavePropagationLaboratory,Boulder,CO 80303) It waswell known that both the ray methodand the PE methodcan handle wave propagationproblemsin 3-D environments,but both of themareapproximations of the waveequation.In ray method,the diffraction termwasdiscardedsoit isgoodfor highfrequency;in PE method,the term of second-order derivationon rangewasdiscardedsoit is goodfor narrow angle.The accumulatedphaseerror restrictsthe rangewithin whicheachmethodis valid.Onehasto choosea bettermethodto adapta $44 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 stabilityand consistency boundsare knownthat imply corresponding ratesof convergence. It isshown,in somecases, howthisinformation can beimprovedto yieldexplicitboundsfor thetruncationerrorbetweenthe exactsolutionof a continuousproblemand the approximatingsolution generatedby the computationalscheme.In particular,thoseboundsare demonstrated for an implicitfinitedifference schemein threedimensions. 11:30 S10. Three-dimensionalunderwater acoustic propagationthrough a steady shear flow. J. S. Robertson (Department of Mathematics, U.S.M.A., West Point, NY 10996-1786), M. J. Jacobson, and W. L. Siegmann (Department of Mathematical Sciences, Rensselaer PolytechnicInstitute,Troy, NY 12180-3590) Previousstudiesby theauthorshaveincorporated currenteffectsinto parabolicapproximations,whenthe currentvectorlay within the vertical source-receiver plane.Here, the currentand the source-receiver plane are not constrainedto be parallel, and thus a solutionto a fully three- dimensional problemisobtained.A classof sound-speed profilesisexaminedthat is advectedin a steady,depth-dependent oceancurrent.A cw sourceis submerged betweena horizontalpressure-release surfaceand a horizontalbottom.Whencertainasymptoticorderconditionson derivativesof thecurrentand soundspeedaresatisfied, parabolicapproximations of canonical form result. Furthermore, if additional restrictionson azimuthalderivativesapply,the azimuthalmodesare decoupledso that the azimuthalangleappearsonly as a parameterin the parabolicequations.Thus the pressurefield may be calculatedin decoupledvertical planes,i.e., in an N X 2-D approximation.Severalexamplesarediscussed, andtheacousticaleffectsof currentstructureareillustrated,with emphasison azimuthaldependence of relativeintensity.[Work supported by ONR. ] 114thMeeting:Acoustical Societyof America $44 WEDNESDAY MORNING, 18 NOVEMBER 1987 POINCIANA ROOM, 9:00A.M. TO 12:00NOON SessionT. ArchitecturalAcousticsIII: Analysis;EducationalFacilities Ewart A. Wetherill, Cochairman F•/ilson lhrig & Associates, Inc., 5776Broadway, Oakland,California94618 RichardM. Guernsey,Cochairman CedarKnolls•4coustical Laboratory, 9 SaddleRoad,CedarKnolls,NewJersey 07927 Chairman's Introduction--9:00 Invited Paper T1. Calculation of free-field deviation in an anechoic room.Ji-qingWang(instituteof Acoustics, Tongji University, Shanghai, People's Republicof China200092)andCai Biao(Jiaotong Universityof Eastern China,Nanchang,People'sRepublicof China) After a reviewof variouscalculationmethodscurrentlyavailablefor free-fielddeviation(in dB) in an anechoic room,theauthors present, inthispaper,a moreprecise andrationalcalculation ofthesoundfieldby solving waveequations in consideration of theinterference of firstreflections only.In thecaseof puretone,the phasefactorplaysan importantrolein the calculation;therefore,the free-fielddeviation(in dB) calculated alsodepends onthesource position, measuring direction anddistance fromthesource, andthefrequency used in thetest,besides otherfactorssuchasthedimensions oftheroomandsoundabsorption characteristics of all theboundary surfaces. For a broadband noise,thecalculation becomes simplersincethephasefactorcanbe ignored, anda general equation ofsucha calculation isgivenforanysource position anddifferent combinations of boundary absorption. Theresults of thecalculation canbeshownin a diagramwiththeaidofa microcomputer,and,hence, a three-dimensional free-field scope wouldbeveryusefulasa guideline todesign andtouse an anechoic room. ContributedPapers 9:35 T2. A graphicrepresentation of acoustics t•ing ray tracing.Philip R. Thompson(Departmentof Architecture,Massachusetts Instituteof Technology, Cambridge, MA 02139) Thisresearch detailsthedevelopment of a computermodelthatdisplaysthe propagationand reverberationof soundwithin an architectural enclosure. Thevisualrendering of acoustics isdoneby simulating sound fieldsusinggraphicalray methods. Unlikemanyotheracoustical raytracingmethods, in thisconcept, raysareemittedfroma receiving point andtracethepathstothesoundsource soastoproduce a pictureofwhata listener hears.Thisthree-dimensional reeursive ray-tracing algorithmhas beencombinedwith an acoustical model,utilizingspecularandfirst-orderdiffuse reflections, toproduce animage(or series ofimages ) depicting the locationsof real andvirtualsources, the time delaysin whichthey arrive,andtheirapproximate energylevels.The free-pathlengthsof the raysaresortedintoframesthatproduce an animationsequence demonstratingthepropagation of a soundfieldtowardsthelistener.The sources' emittedenergyandsubsequent attenuations alonga ray'spathresultin an intensity,whichis thendepicted asa colorat a pixelon the computer's monitortocompose animage.Whiletheprogramismeantto runonmost anymachine, a workstation ispreferred sincea relatively largeamountof computation (involving hundreds of thousands ofrays) isrequired. The purpose of thisanalysis toolis to elucidateandverifyour perceptions of room acoustics throughthe mostreadilyunderstandable visualmeans. S45 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 [Work supported by MIT ProjectAthenathroughthe ComputerResourceLab in theSchoolof ArchitectureandPlanningandby theRobert B. NewmanMemorialFellowship.] 9:50 T3. Effect of flaidel--stracturaicouplingon soundwavesin an enclosure in low-frequencyrange.Pan Jieand D.A. Bies(Universityof Adelaideo Departmentof MechanicalEngineering, GPO Box No. 498, Adelaide, S.A. 5001, Australia) The acousticpropertiesof individualnormalmodesin a rectangular panel-cavitysystemwereinvestigated experimentally.The systemconsisteda heavy five-sidedconcretebox closedon the sixth side with a test panel.Experimentalevidencein thispaperrevealsthedependence of the modaldecaytimeof soundwavesin theenclosure uponthemodalcharacteristicsof the testpanel,suchas the modalcouplingfactors,resonance frequencies, andmodaldecaytimesofthetestpanel.Theresults ofexperimentalmeasurements are usedto verifysomeof the previoustheoretical predictions. The agreement betweenthemeasured andcalculatedacousticalmodaldecaytimeshowsthat themodalcouplingtheorycanbeusedto evaluate thc sound wave behavior in those enclosures,in which the cou- plingbetweenfluid borneand structureborneacousticalwavesis modal and wherethe classicalmodelbasedon the locallyreactiveboundary assumptiondoesnot apply. The phenomenonof the maximumsound energyabsorptionby the panelwasalsoobservedin the experiment. 114thMeeting:Acoustical Societyof America S45 10:05 10:20 T4. Effect of fluldal-structural couplingon the acousticaldecaysin a reverberation room in high-frequencyrange, Pan Jie and D. A. Bies (Universityof Adelaide,Departmentof MechanicalEngineering,GPO TS. An indirect boundary integral equationmethod for architectural acoustics.D. K. Holger (Department of EngineeringScienceand Mechanics,2019 Black EngineeringBuilding, Iowa State University, Box No. 498, Adelaide, S.A. 5001, Australia) Ames, IA 50011) The effectof thecouplingbetweena soundfieldandtestpanelsupon theirdecaybehavioris investigated in high-frequency range.Quasitransientsolutions of theaverage energies in theroomandthepanelsareused A numerical methodandtheresulting computer programfor predictionofinterioracoustic fieldsaredescribed. The methodisbaseduponan indirectboundaryintegralequationformulationof theHelmholtzequation usinghybridlayer potentials.The computerprogramincorporates generalboundaryconditionsdirectly, and utilizespiecewiseconstant variableapproximations. It is capableof predictingthe steady-state response of enclosures with finiteimpedance boundaries, vibratingboundaries,andmultipleinteriorsources. Numericalresultsarepresented for theaeonsticfieldinsidea rectangularprismundervariousboundaryconditionsandwith variousexcitations.Apparentlimitationsof the current programarediscussed, andvariouspotentialapplications in architectural acoustics andnoisecontrolare suggested. [Work performedat the Institutt for Teleteknikk,NorgesTekniskeHogskole,Trondheim,Norway and supportedby NTNF. ] to estimatethe reverberationtimesin the room and to calculatethe decay curvesofa testpanel.Experiments areconducted to verifythesesolutions. The estimatedand measuredresultsare in goodagreementin high-fre- quencyrange,which indicatesthat the averagedecaybehaviorof the roomand panelscan be analyzedin termsof the averagemodalparameters.The amalgamationof the averagemodal parameters(of the test panelsand the reverberation room) canbe relatedto the Sabineabsorptioncoefficient as•bof thepanels.Thisresultprovidesa possible interpretation for the as•b of a modallyreactivesurfaceand suggests a way to estimatethis quantityby statisticalenergyanalysis(SEA) method.The dependence of the as•buponthe roomproperties,in whichthe panelis measured,is demonstrated bothexperimentally and analytically. 10:35-10:45 Break 10:45 T6. Acoustical considerationsin designof semienclosedsmall arms training facility. Felix Sachs,Nelson Lewis, and GeorgeLuz (BioAcousticsDivision, U.S. Army EnvironmentalHygiene Agency, School.In additionto classrooms, thebuildingincludesscreening rooms, edit rooms,a recordingstudio,an AM/FM radiostation,two television studios,and a soundstage.The designapproachis describedand the completedresultsare analyzed. AberdeenProvingGround, MD 21010-5422) Semienclosed smallarmstrainingrangescombinebullettrapfeatures of an indoorrangewith theventilationof outdoorranges.Acousticreverberationof indoorrangesanddirectivityeffectsof outdoorrangesarealso combinedin the semienclosed range.Whendesigners of a semienclosed 11:30 rangefailedtodesign fortheacoustic directivity ofguns, community noise levelsincreased andresidents whohadadaptedto a long-standing noise T9. RASTI for speechisolationevaluation.Klaus Hojbjerg(Brueland Kjaer Instruments, Inc., 185ForestStreet,Marlborough,MA 01752) environment fromanoutdoorrangebegancomplaining. Personnel onthe firingline alsonoticedan increase.Measurements at the firingline demonstrateda sevenfold increasein the occupational noisedose,whilemea- In openoffices,thereis a demandfor speechisolation.As it becomes moreimportantto avoidsurveillance from outsideconference rooms,a reliablemethodof evaluatingthespeechsecuritybecomes necessary. Re- surements made out to 500 m demonstrated an 8-dB increase in the com- munitynoiseexposure. The 8-ribincreasewasattributedto thereflection of theforwardweaponreportfromthebullettrapbaffles. The increase in noiseexposureat the firing line wasattributedto reverberationfrom the sidewalls and roof of the semienclosed range.The measuresto reduce boththe occupational and communityexposures are addressed. search has shown that transmission loss or the articulation index method doesnotgiveenoughinformation. Thisisbecause speech intelligibility, or ratherthelackof speech intelligibility,depends on morethanthesignalto-noise ratio. The RASTI method takes into account the effects of reflec- tions,reverberation,and noise.By usingthe low end of the RASTI scale, thelackof speech intelligibility canbedetermined. Thispaperdescribes the measurement, usingelevatedsignallevels,as well as measurement results. 11:00 T7. Acoustical designfor theMerkin ConcertHall andtheHebrewArts School.PeterJ. George(PeterGeorgeAssociates, Inc.,34 West17th Street,New York, NY 10011) 11:45 Sound control issues werecritical inthedesign ofthemultistory build- TI0. Controllingthe shapeof an aeonsticspectrumautomaticallyin ingthatrequiredrelatively light-weight construction. The AnneGoodmanRecitalHall waslocateddirectlyabovethe Metkin ConcertHall. Dance studios wereadjacent toPianoClassrooms, whichwereadjacent to Practice Rooms. Comparison ismadebetween thetechniques used onthis buildingandthosefortheneighboring JuilliardSchool. 11:15 TS. Acousticaldesignfor theTischSchoolof the Arts. PeterJ. George (Peter George Associates,Inc., 34 West 17th Street, New York, NY 10011) New York University'sTischSchoolof the Arts hasmovedinto a 12storybuildingthat wasguttedandrecycledto providea newhomefor the S45 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 vaeiousSlmeea.Thor Carlsen and Richard J. Peppin (Scantck Inc./ NorwegianElectronicsa/s, 51 Monroe Street,Suite 1603,Rockville, MD 20850) Often,whena roomis excitedby a specified spectrum,the resulting spectrumdiffersin shapefrom that desired.Causedby the impedanceof the spaceand by the objectswithin, in the past,the changehasbeen accounted forbytheuseof a spectrum shaperbeforethenoisegenerator, whichwasmanuallyadjusted sothat the desiredshapeandlevelof the outputwasachieved.This paperpresents a discussion of a bus-controlled spectrumshaperthat, in conjunctionwith a computercontrolledrealtime analyzer,allowsthe shapeof the desiredspectrumto be achieved withoutoperatorintervention.Usesfor vibrationexcitationwill alsobe discussed. 114thMeeting:AcousticalSocietyof America S45 WEDNESDAY MORNING, 18 NOVEMBER 1987 TUTTLE SOUTH ROOM, 9:00 TO 11:00 A.M. SessionU. Noise III: Alert and Notification Systems T. JamesDuBois, Chairman SouthernCalifornia Edison,2131 Walnut GrooeAoenue,Rosemead,California 91770 Chairman's Introduction--9:00 Invited Papers 9:05 U1. The FederalEmergencyManagementAgency'salert and notificationsystemsreviewi•rogram.Craig S. Wingo (FederalEmergencyManagementAgency,Washington,DC 20472) Nuclear powerplantsthroughoutthe United Statesare requiredby the Nuclear RegulatoryCommission (NRC) to providethe meansfor earlynotificationandclearinstructionof thepopulacewithinthe 10-mile radius EmergencyPlanningZone (EPZ) around each plant. The regulationsand the Federal Emergency ManagementAgency's(FEMA) technicalprogramfor reviewingand approvingalert and notificationsys- temsare discussed in this paper.The FEMA programconsists of an acoustical/engineering reviewof the systemdesign,coupledwith a telephonesurveyof residentsof the EPZ followingactivationof the systemto determineitseffectiveness. To date,systemdesignshavebeensubmittedfor everynuclearpowerplant in the nation.The engineering designreviewhasbeencompletedat 68 sites,anda telephone surveyhasbeencompleted at 63 sites.Recently,the technicalstandardsagainstwhichalert and notificationsystemsare reviewedhave come under scrutiny inAtomic Safety andLicensing Board litigation andFEMAhasdefended theadequacy of thecurrentstandards. Effortsareunderwayto placetheOutdoorSoundPropagation Modelusedin thealert and notificationsystemreviewsinto FEMA's IntegratedEmergencyManagementInformationSystemfor applicationto civil defenseprograms,hazardousmaterialsprogram,and other programsin and outsideof FEMA. 9:30 U2. Designof alert and notificationsystems.Louis C. Sutherland (Wyle Laboratories,128 Maryland Street, El Segundo,CA 90245) and Eric Stusnick(Wyle Laboratories,2001 JeffersonDavis Highway, Arlington, VA 22202) Alert and notification systems arerequired bytheNuclear Regulatory Commission around nuclear power plantsto providea meansof alertingthepublicwithina 10-mileradiusof a possible emergency conditionthat may requireevacuation.The mostcommonmethodof accomplishing this is throughthe useof audiblesignals from high-poweredsirens.The basicproblemsinvolvedin the designof suchsystemsare discussed. These problemsincludespecification and validationof the sirenoutputsignals,the predictionand validationof the sirensignalas receivedin the community,and nonacoustic factorssuchas sireninstallationand control. Generalcharacteristics of currentlyavailablesirensystems--bothelectromechanical and purelyelectronic types---arealsobrieflyreviewed.Statisticalaspectsof short-term(secondto second)and long-term(day to day) variations in sirensignallevelsreceived in thecommunityarediscussed, andthisvariationin thewarning signallevelis evaluatedwith respectto overallsystemeffectiveness. Othermeansof conveying thewarning, includingthepresentation of alteringmessages overautomaticallyenergizedradiosandsiren-equipped emergencyvehicles,are alsobriefly considered. 9:55 U3. Evaluationof sirenalert notificationsystems.Van M. Lee (Analysis& Computing,Inc., Hicksville,NY 11802) This paperdescribesthe essentialelementsin evaluatingthe effectiveness of a large outdoorfixed siren warningsystem,thestate-of-the-art of thecomponent evaluationprocesses, andthepotentialimprovements in the future.The essentialcomponents in evaluatingthe coverage--thepercentageof populationwithin the planningzonealerted-•consistof: sirenacousticperformance(dB versusfrequencyversusdirectivity), site terrainandmeteorological effectsonsoundpropagation, populationstatistics(activitypatternsandagedistributionby household size), buildingattenuationeffects,and activity-soundlevelresponse relationships. The lack of standardtest procedures,whether laboratoryor field, and data reportingformatshas led to many disputesconcerningomnidirectionaland directionalsirens.Standardmethodsfor site qualification,siren testing,and data reportingsimilar to ANSI S1.34 (1980) and S1.35 (1979) are discussed in relationto their effecton thedesignandevaluationof sirensystems. Also discussed arethe effectsof air absorption,wind and temperaturegradients,operatingcharacteristics (omnidirectionalor rotating), and terraineffects,including Rasmussen's applicationof barrierdiffractionandimpedancegroundeffects.By correlatingthephysicalsiren S47 J. Acoust.Soc.Am.Suppl.1, Vol. 82, Fall 1987 114thMeeting:AcousticalSocietyof America S47 soundlevelcoverage tothesocialresponse levelof thealertedpopulation, a systematic methodology ispresented, which evaluatesthe overalleffectiveness of a largeoutdoorsirenalert system. 10:20 U4. Outdoor sirens:How do they work for peopleindoors?David N. Keast (HMM Associates,inc., 336 Baker Avenue, Concord, MA 01742) Peopleare normallyindoorswherethe soundsfrom outdoorwarningsirensare greatlyattenuated by buildingstructures. Nevertheless, practicalexperience andthe resultsof testsby the FederalEmergency Management Agencyandothershaveshownthat sirensounds arequiteeffective for alertinglargepopulations-including thoseindoors. Whyisthisso?Based upondatain theliterature, threepossible explanations areexamined in thispaper:( l ) thepossibility thatthesounds fromsirensoutdoors, eventhoughattenuated throughbuilding structures, arestillsufficient to directlyalertpeople indoors; (2) thepossibility thatpeople indoorsaremoreresponsive to sirensounds thanpeopleoutdoors; and(3) thepossibility thatpeopleindoors arealerted byotherpeople whoareoutdoors whentheyhearthesirensounds. Thisexamination indicates that all threemechanisms couldbecontributing to theeffectiveness of sirensounds indoors. Theirrelativesignificanceisnotknown,butprobablyvarieswithgeographic location,weatherconditions, andtimeof day. ContributedPaper 10:45 sign,theimportance of finite-amplitude effectshasbeenneglected in the US. The influenceof high-intensitysoundson the performanceof air warningsirens.FredericG. Pla (SverdrupTechnology,Inc., NASA LeRC Group, 16530 CommerceCourt, P.O. Box 30650, Midpark Branch,MiddleburgHeights,OH 44130) recentpast,eventhougha reviewof earlierworksshowsthat researchers suchasReyleigh,in 1903,King,in 1919,andlater,Hart, in 1926,noticed the extra acousticlossesassociated with high-intensity soundsources. Comparisons between linearandnonlinear models ofwavepropagation in a sirenhornillustratethe importanceof thesetypesof losses, andshow how theycanbe minimized.Acousticsaturation,whichlimits the sound- Theuseofsirens asairwarning devices hasbeenwidespread forseveral hundredyears.Due to theveryhighsound-pressure levelsgenerated, finite-amplitude effectscanresultin largelosses leadingto poorsound high-intensity soundsource, andtheimplications of finite-amplitude ef- generation efficiency. Althoughmuchwork hasbeendoneon sirende- ment are also discussed. WEDNESDAY MORNING, 18 NOVEMBER 1987 pressure levelthatcanbeperceived bya receiverat a givendistance froma fectsontheratingofhigh-intensity soundsources in a laboratoryenviron- TUTTLE CENTER ROOM, 9:00 TO 11:50 A.M. SessionV. PhysicalAcousticsIll: Scattering,Diffraction, and Radiation Ronald A. Roy, Chairman Departmentof MechanicalEngineering,MasonLaboratory,Yale University,9 HillhouseAvenue,New Haven, Connecticut 06520 Chairman's Introduction--9:00 ContributedPapers 9:05 V1. Monostatic angular distributionsand their relation to resonancesin acoustical scattering from fluid loaded solid elastic objects. G. Gaunaurd(NSWC, White Oak, SilverSpring,MD 20903-5000)and M.F. Werby (NORDA, NSTL, MS 39529) It is possibleto detectresonances from backscattered echoes(form functions)by varyingthe nondimensional frequencykL/2 in smallincre- ments.Thisgivesrisetoa familiarpatternofnullsfollowedbya sharprise for the sphericalcaseand similarresultsfor the spheroidal case.This S48 J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 approach,however,limitsoneto oneaspectanglefor three-dimensional nonspherical problems andoftenentailstedious andtimeconsuming calculationsto findtheoftennarrowresonance regions.Anotherapproach suitable forthree-dimensional problems istoexamine closely spaced monostaticangulardistributions enablingoneto lookfor possible resonances overa continuous rangeof aspectangles.This methodis usedto examine scattering fromsolidsteelspheroidal shells.TheresultsindicatethatRayleightyperesonances areaspectangledependent in contrastto ourrecent resultson spheroidal shells.The resultsareconsistent withthestanding waveinterpretationof resonances proposedby useearlierfor shellsand employedin workon materialpropertiesof solidelasticspheroids. 114th Meeting:AcousticalSocietyof America $48 9:20 V2. Acousticscatteringfrom an elastic prolate spheroidin a shallow- water waveguide.Gary S. Sammelmannand Roger H. Hackman (Physical AcousticsBranch, Naval Coastal SystemsCenter, Panama City, FL 32407) The bistaticscatteringof a broadbandacousticpulsefrom an elastic prolatespheroidin a shallow-water waveguide isstudied.The calculation is basedon the formalsolutionto the scatteringin an inhomogeneouslayeredwaveguidepresentedat a previousmeetingof the AcousticalSociety JR, H. Hackman and G. S, Sammelmann,J. Acoust. Soc. Am. Suppl.1 81, S42 (1987) ]. The waveguide is assumedto beboundedfrom aboveby a soundsoftair-seawaterinterface,and from below,by a rigid sediment-seawater interface.The soundspeedis takento bea monotonically decreasing functionof depth.The emphasisof this studyis on extractingelasticinformation fromthescattered waveform. Theimportance of the rescattering of the acousticpulsebetweenthe boundariesand the targetis alsostudied. due to leaky Lamb wave generation,too). This simpletechniquehas provedto beveryusefulin materialevaluationvia measuring bothsurface and Lamb wavevelocities.On the otherhand,a comprehensive experimentalstudy[M. deBilly, L. Adler, andG. Quentin,J. Acoust.Soc.Am. 72, 1018(1982) ] showedmuchstrongerbackscattered signalsthan predictedlater by theoreticalworksbasedon back reflection.New experimentalresultsare presenteddemonstrating that this back reflectionis negligiblein the backwardsignal,which is simplybackscattering from inevitablematerialinhomogeneities greatlyenhanced by surfaceor Lamb wavegeneration.Therefore,besidemeasuringmacroscopic elasticpropertiesin the previouslysuggestedway, the backscatteredsignalcan be usedtocharacterize surfaceandnear-surface materialinhomogeneities as well. [Thisworkwassupported by theU.S. Departmentof Energy,Basic EnergySciences, Grant DE-FG02-84ER45057.A004.] 10:20 V6. Measurements of Rayleigh wave scattering at a corner. Peter B. Nagy and Laszlo Adler (Departmentof WeldingEngineering, The Ohio StateUniversity,Columbus,OH 43210) 9:35 V3. Acoustic scattering from large aspect ratio, axisymmetric shells. Roger H. Hackman and Douglas G. Todoroff (Physical Acoustics Branch,Naval CoastalSystemsCenter,PanamaCity, FL 32407) The spheroidal-coordinate-based transitionmatrixhasbeensuccessfully appliedto the solutionand the analysisof the acousticscattering from largeaspectratio solids[R. H. Hackmanand D. G. Todoroff,J. Acoust.Soc.Am. 78, 1058-1071 (1985)]. However, this approachrequiredspecialcaredueto the existence of elasticstatesthat wereweakly coupledto theexternalacoustic field.Here,a newversionof thespheroidal formalismis presented,onewith improvedstabilityand which is better suitedto thin shells.Thisapproachisappliedto theacousticscattering froma largeaspectratio,elasticprolatespheroidal shellof constantthickness.Theoreticalpredictionsare comparedwith experimentalmeasurements,anda physicalexplanationis givenfor the moreprominentfea- Theproblemof Rayleighwavescattering in a homogeneous, isotropic wedgeby experimental meanswasinvestigated. For anincidentRayleigh wave,reflectedandtransmittedRayleighwavesare generated,aswell as longitudinal andtransverse cylindrical bulkwavesinitiatedbythevirtual sourceat thecornerline.Sector-shaped testpieceswereused,whichfocusedbackthedivergingbulkwavesto thesamecornerlineafternormal reflectionat the cylindricalsurface.By this simplegeometry,both the phaseandamplitudeof all fourcomponents wereableto be measured. Experimental results arepresented to showthevaryingenergypartition amongthese modes asafunction ofwedge anglefrom90øto 150øinaluminum.Furthermore,theRayleighwavereflection andtransmission results arecompared to recenttheoretical predictions by Gautesen[ WaveMotion 9, 51-59 (1987)]. [This work was supportedby the U.S. Depart- ment of Energy, Basic Energy Sciences, Grant DE-FG0284ER45057.A004. ] tures in the backscattered formfunction. 10:35 9:50 V4. Comparisonof quadraticand piecewise-constant boundaryelement solutionsof the surfaceHelmholtz equation.JosephJ. Shirron (Sachs/ Freeman Associates,Landover, MD 20785) A comparison of the accuracyandcomputationtime of two different boundaryelementsolutiontechniques for thesurfaceHelmholtzequation of a rigidscattererispresented. In thesimplest case,thebody'ssurfaceis represented by an assemblage of planartriangles,overwhichthesurface pressure istakentobeconstant. In themoresophisticated case,quadratic- isoparametric elements areused.Results arepresented forthecomparison of ratesof convergence withmeshsizeandCPU timesfor thetwo methods.The effectsof varyingthe orderof Gaussianquadratureare shown. Furtherresultsaregivencomparing theefficiency of Schenck's algorithm for eliminatingthenonuniqueness problemat thebody'sinterioreigenvalueswith the modifiedGreen'sfunctiontechniqueof D. S. Jones.The quadratie-isoparametric solutionmethodtogetherwith Schenck's algorithm isfoundto bethe moreei•cientmeansof obtainingan approximate solutionto thesurfaceHelmholtzequation.[ Work supported by Naval ResearchLaboratory,Washington,DC. ] V7. Farfieldpatternsof rigid nonaxiallysymmetricobstacles: Example calculationsusingthe Watermanschemefor approximationof the T matrix. Angie Sarkissian(Sachs/Freeman Associates, Landover,MD 20785) and Allan G. Dallas (Code 5131, Naval ResearchLaboratory, Washington,DC 20375-5000) To our knowledge, therehavebeenno reportsof numericalapplicationsof the Watermanschemeto rigid scattersof nonaxiallysymmetric geometry. Here,theresults ofcomputations for variousslender obstacles of thissortaredescribed. Emphasis is placedupontheamountof work required forsuchcalculations (relativetothatnecessary foraxiallysymmetricshapes) anduponspecific details, suchasthenumbers ofspherical wavesand quadraturepointsrequired. 10:50 VS. The effectof finite surfaceacousticimpedanceon soundfieldsneara smooth diffracting ridge. Ji-xun Zhou, James A. Kearns, Yves H. Berthelot,and Allan D. Pierce (Schoolof MechanicalEngineering, GeorgiaInstituteof Technology, Atlanta,GA 30332) 10:05 VS. Another look at the Rayleigh angle backscattering.Peter B. Nagy and Laszlo Adler (Department of Welding Engineering,The Ohio State University, Columbus,OH 43210) The incresedbackscattering from a liquid-solidinterfaceat the Rayleigh angleis a well-knownindicationof leaky surfacewave generation (similarphenomena wereobserved from thin platesimmersedin liquid $49 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 Laboratoryscaleexperiments usingsparksources arebeingconducted to studylong-range propagation of soundoverhillsandvalleys.The effectof thefiniteimpedance of thediffraeting scaledhill on theacoustic fieldisinvestigated, witheitherplywoodor plywoodcovered withcarpet. Particularattentionis givento the caseof smallgrazinganglesof incidenceof the soundwaveon the diffractingsurface.The modelof Delany andBazleyis usedto determinethe impedance of thematerialfrom the valueof the effectiveflow resistivitya. The insertionlossof the scaled ridge isinferred frommeasurements oftheratiooftheFourier transforms of thediffractedpressure fieldto that of the freefield,with subsequent 114thMeeting:Acoustical Societyof America S40 digitalsignalprocessing performed in thefrequency domain.Experimental dataare comparedwith thetheoryof matchedasymptoticexpansions [J. Acoust.Soc.Am. Suppl. I 79, S30 (1986) ] for situationswherethe diffracted acoustic pressure ismeasured eitheralongtheridge,oralonga verticalaxisat theapexor behindtheridge.[Worksupported by NASA, LangleyResearchCenter.] Soundradiationby finitecylindrical stiflened shells,in lightor heavy fluid,isstudiedusingmodalcalculation.Numericalresultsonpowerradiation,shellquadraticvelocity,and radiationfactorare given,and the influenceof parameterslike stiffenersdistribution,stiffenerand shell damping, typeofexcitation, etc.,isdiscussed. Finallytheanalysis ofshell modesthat contributemostto the radiationis presented. It isshown,in particular,that modeswhosestructurallossesequalradiationlosses dominate the powerradiation, whatever the fluid (however, for light 11:05 V9. Diffractionof pulsesat a sharpedge.Gerrit Schouten(Aerospace Department,Delft Universityof Technology,Kluyverweg1, 2629 HS Delft, The Netherlands) fluids,thesemodeshavehighradiationefficiency andlow radiationefficiencyfor heavyfluids).Thisanalysis allowstheunderstanding of radiationcurvesingularities, andshowsthatcriticalfrequency phenomena arecompletely differentfor heavyandlightfluids.[Work supported by the Minist•re de la D•fenseContratD.R.E.T.N ø85-065.] The closedexpression for the diffractedfieldof a soundpulseis derived in a straightforwardmannerusingSommerfeld'smethodfor the generationof many-valuedsolutionsfrom knownone-valuedsolutionsof thewaveequation.The sphericalprimarywavefrontcarriesa delta-func- tion disturbance, whereasthe spindle-shaped diffractionfront carriesa square-root singularityboundingthe filleddiffractionregionwheretwo dimensional aspects aredominating. Integration w.r.t.timeoffiringofthe pulseyieldsthe transientbehaviorof a hydrodynamic switch-onsource witha unit-step wavefront.Thewell-known resultof Cagniard( 1935) is thusreproducedin a directway. Plotsof the velocityfield insidethe diffractionregionshowthe peculiarities of the transitionfrom the wave solutiontothesteadyLaplacesolution.Applyingthemethodofdescent to thethree-dimensional solutions yieldstheknowntwo-dimensional equivalents,againin a moredirectwaythanknownfromtheliterature[R. D. Turner( 1956);H. A. Lauwerier( 1962);E. M. deJager(1964)]. 11:35 Vll. On the acousticexcitationsof an elongatedelasticsolid. Kevin L. Williams,Douglas G. Todoroff,andRogerH. Hackman(Physical Acoustics Branch,Naval CoastalSystems Center,PanamaCity, FL 32407) Previously, an analysisof the acoustically coupled,axisymmetric, elasticexcitations ofa fluid-loaded, prolatespheroidal targetwaspresented [D. H. Trivett, G. S. Sammelmann, and R. H. Hackman,J. Acoust. 11:20 Soc.Am. Suppl.181,S13(1987)]. At thismeeting, a controversy arose VIO. Acoustic power radiation by cylindrical stiflened shells. asto thephase speed oftheelastic excitations. Here,directexperimental B. Laulagnet andJ. L. Guyader(Laboratoire Vibrations-Acoustique, andtheoretical evidence ispresented thatsupports ourcontention thatthe InstitutNationaldesSciences Appliquresde Lyon,69621Villeurbanne velocity ofpropagation oftheelastic waves thatpropagate onanelongatCedex, France) edsolidtargetat low frequencies is essentially thebarspeed. WEDNESDAY MORNING, 18 NOVEMBER 1987 TUTTLE NORTH ROOM, 9:00 TO 11:50A.M. SessionW. StructuralAcousticsand Vibration III: Analysisand Measurementof StructuralVibrations Joel M. Garrelick, Chairman Cambridge •lcoustical •lssociates, Inc., 54 RindgeAvenueExtension, Cambridge, Massachusetts 02140 Chairman's Introduction--9:00 ContributedPapers 9:05 WI. Impact testing of light objects. Mohamed K. Abdelhamid (Production EngineeringDepartment, Helwan University, Helwan, Cairo,Egypt) andAnna L. Pate (Departmentof Engineering Science and Mechanics,Iowa StateUniversity,Ames,IA 50011) Thispaperinvestigates thedesignof a measurement systemthatmea- sures theimpactsignature ofa lightobject. Thedesign isbased ona freely supported piezoelectric forcetransducer. Effectsof top mass,basemass, andstiffness ofthetransducer wereinvestigated. Twomodels wereusedin the analysisof the impact,a linearmodelwith a 3 degreeof freedom systemand a Hertzian model.Using the HertzJanmodel,a criterionfor choosing theoptimumsizeof thebasemasswasdeveloped. Thiscriterion is basedon the least-square errorin the forcespectrum.Similarresults wereobtained bythelinearmodelandnumerical simulations. Experimentalconfirmation of theresults wasconducted usingsmallTeflonandny- $50 J. Acoust.Soc.Am.Suppl.1,Vol.82, Fall1987 lonballsanda piezoelectric forcetransducer. Theresultsagreedwellwith the modelpredictions. 9:20 W2. Wave effectcorrectionto reciprocitycalibrationresultsof vibration primary standards. Li-Feng Ge (National Bureau of Standards, Gaithersburg, MD 20899andAnhiBureauofStandards andMetrology, Hefei, People'sRepublicof China) Thepaperrevealsthestress waveeffect(SWE) onreciprocity calibration resultsof vibrationprimarystandardsandderivesexactcorrectformulasof mechanical standards according to reciprocitytheoryandstress wavetheory.The paperfurtherexpounds that boththe electrodynamic reciprocitymethod (ERM) and the piezoelectric reciprocitymethod 114thMeeting: Acoustical SocietyofAmerica S50 10:20 (PRM) needthiscorrection at higherfrequencies andshows thattaking thismeasure caneliminate asystem errorof10-2to10 3ordersofmagnitude. 9:35 W3. Viscoelasticeffectsonsolidamplitudeandphasetransformers.L. M. B.C. Campos (Instituto Superior Trcnico, 1096 Lisboa Codex, Portugal) Thelongitudinal oscillations of nonuniformbarsareusedasdisplacementamplifiers in powertoolsandotherdevices. Theseoftenoperatenear resonance[ E. Eisner,J. Acoust.Soc.Am. 35, 1367-1372 (1963) ], sothat the elasticmodelshouldbe replacedby a viscoelastic one,whichincludes damping.In the presentpaper,the longitudinalvibrationsof a tapered viscoelastic bar are discussed generallyfrom the waveequationsfor the displacement andstrain.Exactsolutions areobtainedfor theexponential, catenoidal, sinusoidal, andinverseshapes, andalsofor theGaussian and power-lawshapes.These solutionsgeneralizeearlier resultsfor elastic bars;e.g.,the elasticGaussianbar [D. A. Bies,J. Acoust.Soc.Am. 34, 1567-1569 (1962)] is generalizedto a viscoelastic one. Diagramsof wavenumber anddampingratio versusfrequencyand viscousrelaxation time are presentedfor severalshapesof taperedbar; they describethe propagation anddissipation of oscillations andtheir effectson amplitude and phase. 9:50 W6. Vibration analysis of mass-loaded beams. Dhanesh N. Manikanahally and Malcolm J. Crocker (Department of MechanicalEngineering,Auburn University,Auburn, AL 36849) A generalprocedure fordetermining thedynamicresponse of a massloadedfreefreebeamsubjected to a harmonicandtransientforceisgiven. Thoughfreefreebeamsare considered for analysis,the sameprocedure could be extended for other end conditions also. The beam is assumed to havestructuraldampingfor determiningthe steady-state response due to harmonicforceexcitation.The modeshapesfor freevibration,dynamic response,and dynamicstraindue to forcedexcitationare presentedin a graphical form.Theanalysis isusedtostudya spacestructure, modeledas a mass-loaded free free beam,by making an exhaustiveoptimization searchfor minimumdynamicresponse due to harmonicand transient excitationforces.The computerprogramdevelopedfor the analysisis usedto checksomesimplebeamproblems[G. B. Warburton, The DynarnicalBehaoiourofStructures (Pergamon,New York, 1976)]. [Work supportedby SDIO/DNA, ContractNo. DNA-001-85-C-0183.] 10:35 W7. Finite element analysis of a large vibrating space structure. B. S. Sridharaand MalcolmJ. Crocker (Departmentof Mechanical Engineering, AuburnUniversity,Auburn,AL 36849) Thelargeflexiblespacestructure to beusedin theStrategic Defense InitiativeProjectis modeledas a long flexiblebeamwith threepoint masses. UsingtheGalerkinmethod,finiteelementequations areformulatedthat takeadvantageof thefactthat thenaturalboundaryconditions comeout as a resultof integrationby parts.Hermite polynomialshave beenchosen for theshapeor interpolation function.A tubularbeammade ofgraphite epoxymaterial,100minlengthwith2.0-mo.d.and1.95-mi.d. W4. Relationship between the impedances and the reflection and transmission coefficientsof a junction. L. J. Maga and G. Maidanik (David TaylorNaval ShipResearchandDevelopmentCenter,Bethesda, MD 20084) A junctionconstitutesthe couplingbetweendynamicsystems.In this case,the one-dimensional dynamicsystemssocoupledare characterized by theirwavenumbers andmassdensities.The interfaceimpedances of the dynamicsystems at thejunctionmay thenbedefined.Thejunctionmay be definedin termsof ajunctionimpedance matrix.The natureof thismatrix will be discussed. This matrix, togetherwith the interfaceimpedances of the dynamicsystems,may thenbe employedto derivethejunction interaction matrix, which consistsof reflection and transmission coefficients. The manner and results of this derivation will be discussed and a few is considered for thepurpose of thisanalysis. Massesof 500, 10000, and 1000kgareplacedat theleftextremeend,at a distance of 20 m fromthe left extremeend,andat therightextremeendof thebeam,respectively. Naturalfrequencies of thelongflexiblebeamarecalculated usingstandardmethodsandthemodeshapes arealsoplotted.Resultsarealso tainedfor a differentlocationand increasedvaluesof the largestmass. Computercodesare beingpreparedto obtainthe dynamicand steadystateresponse of thebeamsubjected to animpulse anda sinusoidal force. 10:50 WS. Procedure for the measurement of wavenumber/frequency admittanceof structures.Karl Grosh, W. Jack Hughes,and Courthey B. Burroughs(AppliedResearch Laboratory,The Pennsylvania State exampleswill becited. University,P.O. Box 30, StateCollege,PA 16804) 10:05 WS. Damping and vibration analysisof bondedbeamswith a lap joint. Mohan D. Rao and Malcolm J. Crocker (Mechanical Engineering Department,AuburnUniversity,Auburn,AL 36849) In thispaper,a theoreticalmodelto calculatethe lossfactorsand the resonancefrequenciesof flexural vibrationsof a systemof two parallel beamsbondedwith a singlelapjoint by a viscoelastic materialhasbeen A procedure formeasuring thewavenumber/frequency admittance of structures usingan arrayof drivesin a standingwavepatternwasdevelopedin a paperpresented in theASA meetings in Indianapolis, IN on 14 May 1987[K. Grosh,J. H. Hughes,andC. B. Burroughs, I. Acoust.Soc. Am. Suppl.181, S73(1987)]. In thispaper,dataarepresented to verify thevalidityof themeasurement procedure. The wavenumber/frequency spectra arepresented forarraysofpointdrivesandthevibration responses of lightlydampedbeamsand longbeamswith heavilydampedends.The effectof steeringthearrayto differentwavenumbers andvaryingthenumber of drives is examined. developed. First,equations of motionof thejoint regionarederivedusing a differentialelementapproachconsideringthe transversedisplacements of the upperand the lowerbeamto bedifferent.The normalforcebetween eachbeamand the adhesivelayeris represented by a Pasternakbasemodel, whichconsistsof closelyspacedlinear springs.The shearforceat the interfaceismodeledusinga viscousmodelfor friction.The resultingequationsof motion,togetherwith equationsof transverse vibrationsof the beamsawayfrom thejoint, are solvedusingmotioncontinuityconditions and boundaryconditionsat the free endsof the beams.Equationsfor calculatingthe resonancefrequencyand the lossfactor for the caseof clamped-clampedboundaryconditionsare then derived.Numerical resultsaregenerated andarecomparedwith experiments for a systemof two beamswith a simplelapjoint madeof graphiteepoxycompositematerial. [Work supported by NASA-MSFC. ] S51 J. Acoust.Soc. Am.Suppl.1, Vol.82, Fall 1987 11:05 W9. Influenceof variousparameterson the transmissionof vibrational power.T. Gilbert,J. M. Cuschieri,M. McCollurn,andJ.L. Rassineux (Centerfor Acousticsand Vibrations,Departmentof OceanEngineering, Florida Atlantic University,BocaRaton, FL 33431) The transmission of vibrationalpowerbetweentwo thin platesin an Lshapeconfigurationis investigated usingan SEA model.Only bending wavesareconsidered inthemodel.Expressions aredeveloped forboththe ratio of energylevelsin the two platesand the ratio of the transmitted powerto theinputpower.It isfoundthat theseratiosareonlydependent uponthreeparameters: frequency,dampingsof the plates,and coupling lossfactorsbetweenthe two plates.Therefore,a way to decreaseboththe 114thMeeting:AcousticalSocietyof America S51 transmitted powerandtheenergylevelin thereceivingplateovera wide rangeof frequencies is to increasethe dampingof the sourceplate.The couplinglossfactor is influencedby the physicalcharacteristicsof the plates,suchas area,thickness,and material.The effectof theseparam- on the response are analyzed.As a verification,the obtainedresultsare comparedwith finite elementanalysisand statisticalelementanalysis. [Work supportedby NASA.] eterson thecouplinglossfactorandonthepowerandtheenergyratiosis foundto besmall.Thusthemostefficientwayto controlthe transmission of powerand the energylevelsis to increasethe dampingof the plates. This resultis in agreementwith otherresultsobtainedusingthe energy influencecoefficient method.[Work supportedby NASA.] 11:35 W11. Experimental measurementof powertransmissionin an L-shaped beam. M. McCollurn, T. Gilbert, J. Rassineux, and J. Cuschieri (Departmentof OceanEngineering, FloridaAtlantic University,Boca 11:20 W10. Power flow method for an L-shai•ed plate. J. L. Rassineux, J. M. Cuschieri, M. McCollurn, and T. Gilbert (Center for Acoustics and Vibrations,Departmentof Ocean Engineering,Florida Atlantic University,BocaRaton, FL 33431) The powerflow methodis extendedto the analysisof the vibrational powerflow for two-dimensionalL-shapedplatestructures.The analysisis for a platethat is assumed to be simplysupported alongthe edgesand pinnedat thejunction.Structuraldampingis introducedby meansof a complexelasticitymodulus.The transmitted powerfromoneplateto the other is computedusinginput and transfergeneralizedmobilities.The mobilityfunctions areobtainedfromthesolutionfor theresponse of a fiat plateusingeitherpointor edgemomentexcitation.The analyticalsolutionobtainedenables computation of theresponse fora broadband excitation, regardless of the rangeof frequencyconsidered. Differentlocations of theexcitationpointanddifferentvaluesof dampingandtheirinfluence S52 d.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 Raton, FL 33431 ) The vibrationalpowertransmittedacrossthejoint betweentwobeams formingan L-shapedstructure,whenone of the beamsis excitedby a point forceat the free end, is measuredexperimentally.One methodof calculatingthe powertransmission is basedon the assumption that the powertransmittedto thereceiverbeamisequalto thepowerdissipated by that beam. The dampinglossfactor usedin the calculationis obtained frommeasurements onthereceiverbeam.Theresultsof theexperimental analysis arecompared tothoseobtainedusinga powerflowtechnique, the resultsof which havebeenverifiedby the closedform solutionfor the globalstructure.The measurements showverygoodagreementwith the analyticalresultsin termsof the locationof the resonantfrequencies, exceptfor lowfrequencies. The measured levelsare,however,significantly lowerthanthe predictedlevels,especially at theresonantpeaks,indicating that the measureddampinglevelsare underestimated. The measuredlevelsaregenerallyverycloseto thepredictedlevelsawayfrom the naturalfrequencies. [Work supportedby NASA andONR.] 114thMeeting: Acoustical Society ofAmerica S$2 WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 TUTTLE NORTH ROOM, 12:30TO 1:50P.M. SessionX. Educationin AcousticsI: Project Laboratoriesfor Undergraduates Allan D. Pierce, Chairman Schoolof MechanicalEngineering, GeorgiaInstituteof Technology, Atlanta, Georgia30332 Invited Papers 12:30 XI. Developmentof an undergraduate ex•rimental acousticscenter.Jacek Jarzynski(Schoolof Mechanical Engineering,GeorgiaInstituteof Technology,Atlanta, GA 30332) A centerfor experimental acoustics andultrasonics wasrecentlydeveloped for usein a senior-level laboratory coursethat is requiredfor the BSME degreeat GeorgiaTech. The centeris a distinctfacility with permanentlyassigned instrumentation andequipmentavailablefor hands-onperformance of experiments and laboratoryprojectsby students;possibleexperimentsare related to current courseofferingsand on-going researchactivitieswithin theSchoolof MechanicalEngineering.The experimentsthat havebeendevelopedfor thecenterallowthestudents to studysomeof thebasicfeaturesof wavemotionandto learnaboutsometypical applicationsof acoustics.Representative experimentaltasksthat canbe carriedout within the centerinclude: measurements ofpressure levelsandfrequency spectraofsoundfromenvironmental sources; application ofthe ultrasoundto nondestructive testingof materials;and measurement of fluid flow usingan ultrasoundDoppler technique.To the fullestextentpossible,the instrumentation of the centerhas beenchosensuchthat the studentswill be introducedto the lateststate-of-the-art data acquisition,storage,and processing techniques. The objectivesof this type of laboratoryinstructionare outlined,with somesamplestudentcomments.The illustratedtalk will be accompanied by severallive demonstrations of the experimentsthat studentscarry out within the center. 1:10 X2. Project-typeacoustics experiments in the USNA acoustics lab. S.A. Elder (PhysicsDepartment,United StatesNaval Academy,Annapolis,MD 21402) For someyears,the Naval AcademyPhysicsDepartmenthasoffereda senior-levelcoursein physical acoustics, accompanied by a weeklylaboratory.Sincethe classes are small,the laboratoryoffersa chancefor tutoring the studentsin more advancedand creativeprojectsthan the usual "classic"lab experimentsfor coursesat this level.Usuallythereis time to performthreeor four projectsoverthe semester,eachonedefined by a generalareaof interest(suchasenvironmen tal acoustics, architecturalacoustics, computational anddata acquisitionmethods,etc.). Sometimesthe projectspillsover into the next semesterand becomesa credit earningresearch projectbeyondthecourseitself.Examplesof successful studentprojectsare:construction and testingof a loudspeaker(magneplanar,flamedriven, etc.), acousticalinvestigationof a local concerthall or church,construction and testingof an acousticguitar,development of computer-controlled data acquisition and analysissystem.Specificexamples,as well the generalphilosophyof the course,will be given. 1:30 X3. Sonarprojectlaboratoryfor undergraduates. MurrayS. Korman (DepartmentofPhysics,United States Naval Academy,Annapolis,MD 21402) A portablemicrocomputer workstationfor supporting basicacousticexperiments hasbeendeveloped over the pastthreeyearsin the PhysicsDepartmentat the U.S.N.A. Studentsenrolledin underwateracousticsand sonar(a 3-h lecturecoursewithouta laboratory)performfundamentalacoustics experiments in theclassroom withouthavingto utilize largenumbersof analogelectricalboxes.The heartof the workstation(basedon the 6502microprocessor) includesa low-cost,dual-channel,8-bit analog-to-digital board.This workstationfeaturesmenuedsoftwarethat can capturea waveformand displayits trace.Simplemathematicaloperations includesquaringthe waveform,integrating,differentiating,findingthe rms value,and determiningthe probability densityfunction.A 1024-pointFFT programis alsoincludedin the menu.Studentsare not requiredto know computerlanguagesto operatethe workstationand no experimentis automatedor simulated.During nonclass hours,studentsdesignan experimentalprojectthat mightinclude"model"measurements of target strength,beampatternfunctions,reflectionloss,soundvelocity,vibration,and studiesof signalsin noise.The workstations'sversatilitywill be demonstrated. $S3 J. Acoust.Soc.Am.Suppl.1, Vol,82, Fall1987 114thMeeting:Acoustical Societyof America SS3 WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 UM AUDITORIUM, 12:45 TO 3:30 P.M. SessionY. S!•eechCommunicationIV: Analysis and Coding StephenA. Zahorian, Chairman Departmentof ElectricalEngineering,Old DominionUniversity,Norfolk, Virginia23508 ContributedPapers 12:45 YI. Vector quantizationof LPC imrametersbasedon dynamicalfeatures ofhearing.ShigeruOnoandTakashiAraseki(NECCorporation,C&C InformationTechnology Research Laboratories, Kawasaki213,Japan) Newcode-word searchcriteriafor vectorquantization of LPC parametersareproposed. Variousvectorquantization schemes for LPC parametershavebeenproposed previously. In theseschemes, an inputspeech waveformis partitionedinto successive shorttime framesand code-word searchis carriedout underthe criteriabasedon instantaneous speech features.Characteristics of quantizationnoisevary widelyfrom frameto frame,independent of thedynamiccharacteristics of theinputspeech. At verylow bit rates,whereit is difficultto keepthequantization distortion at a low level,variationin the quantization noisesignificantly degrades the qualityof the reconstructred speech."Temporalmasking"experimentsindicatethatthereisa postlyecorrelation beteenthemaskingvalue andthespeech spectral transition rate.Theseexperiments alsosuggest that spectraltransitionratedistortionis emphatically perceived. In this paper,newcriteria,basedon thesetemporalmaskingproperties, areproposed.Theywouldminimizenotonlytheinstantaneous spectraldistortion,butalsothespectraltransitionratedistortion. Usingthesecriteria, vectorquantizationof the LPC parameters is developed. Subjective experiments confirmthat thesecriteriaaresuperiorto thosebasedonlyon instantaneous speechfeatures. articulatorytermsare dueto the simplifyingassumptions underlyingthe analysis model.Usually,the assumption that the systemis all pole [or anto-regressive (AR) ] isthoughttobethemostimportantsimplification. Therefore,it ishopedthattheuseof thelessrestrictivepole-zero[or autoregressive movingaverage(ARMA)] analysismodelwill offer greater opportunities for an artieulatoryinterpretation.A detailedstudyof a numberof differentARMA analysisimplementations hasshownthat the all-poleassumption may not be the mostimportantrestrictionon the articulatory interpretation oftheanalysis results; theassumption ofGaussianexcitationmayproveto beat leastequallyimpeding.[ Research supportedby the Foundationfor SpeechTechnology, fundedby SPIN.] 1:30 Y4. Evaluation ofarticulatory codebooks.$. N. Larar, J. $chroeter,and M.M. Sondhi(AT&T Bell Laboratories,Murray Hill, NJ 07974) Thedesignof a codebook ofarticulatoryshapes wasrecentlyreported [M. M. Sondhi, J.Schroeter, andJ.N. Larar,J.Acoust.Soc.Am.Suppl.1 81,S78( 1987} ]. Thecodebook isusedin anautomaticarticulatoryanalysis/synthesis schemerelyingon goodinitialestimates of vocaltractcrosssectionalareasto startup an optimizationprocedure[J. Schroeter,J. N. Larar, and M. M. $ondhi, Proc. IEEE ICASSP, 308-311 (1987)]. An application-oriented wayto evaluatethecodebookwouldbeto determine 1:00 Y2. LPC voicing periodicity correction. Paul Milenkovic and SotaroSekimoto(WaismanCenter,Universityof Wisconsin-Madison, 1500HighlandAvenue,Madison,WI 53705-2280) Theautocorrelation method oflinearpredictive coding (LPC)analysismatches thefirstp samples of theautocorrelation functionof theim- pulse response ofanall-pole filtertotheautocorrelation function comput- howmuchthe final optimizedshapedeviatesfrom the shapeinitially selected. A moregeneralevaluationis to computetheaveragedistortion incurredin quantizingsomeinputrecordof speech. Usingthisprocedure withanappropriatedistancemeasure,variousaspects ofeodebookformation areevaluated.In particular,the focusis on the strategies utilizedfor generatingtrainingdata. Also consideredis the effectof differentapproachesto code-wordformationoncethe trainingdata havebeenobtained. Evaluation results for this work done to refine the codebook will be presented alongwith the performance asa functionof codebooksize. edfroma speech waveform. Thea•tocorrelation function ofthespeech waveform isinfluenced bytheperiodicity ofthevoice source, aswellasby thewaveform window.Thisinfluence, whenincorporated intotheautocorrelation functionof theall-polefilter,resultsin errorsin formantfrequencyandbandwidthvaluesthatbecomeimportantat thefundamental i 1:45 YS. Speakeradaptationin articulatoryspeechanalysisby synthesis. frequencies typicalof femalespeech. Theuseofananalysis-by-synthesis J. Schroeter,J. N. Larar, andM. M. Sondhi(AT&T Bell Laboratories, technique isproposed, wheretheautocorrelation function oftheoutput waveform ofanLPCsynthesizer ismatched tothespeech dataautocorrelationfunction. Theall-pole filterdetermined byLPCanalysis provides a matchwhenthesynthesizer isoperated at anextremely lowfundamental frequency. Thesynthesizer fundamental frequency is increased in small steps andcorrections aremadetotheall-pole filterateachsteptopreserve theautocorrelation functionmatch.Thisprocess continues untilthecor- rectvalueofthefundamental frequency isreached. [Worksupported by NIH Grant NS 21516-3.] speech analysis as a means for studying articulation. Louis Bores and Johan de Veth (Institute of Phonetics,Nijmegen University,P.O. Box9103, 6500HD Nijmegen,The Netherlands) It is well documented that linearpredictionanalysiswill only allow prediction coefficients to relateto vocaltractshapes underveryspecial conditions, and that the problemsin interpretingLP analysisresultsin S$4 Usingan articulatoryapproachfor low-bitrate speechcoding[J. Schroeter, J. Larar,andM. M. Sondhi,J. Acoust.Soc.Am. Suppl.1 80, S19 (1986); Proc.IEEE ICASSP 308-311 ( 1987) ], meansfor adapting ananalysis/synthesis scheme todifferentspeakers wasinvestigated. Since a crucialfeaturefor thisapproachto speechcodingis the useof an articulatory codebookof tract shapesand relatedtransferfunctions[M. M. Sondhi, J.Schroeter, andJ.N. Larar,J.Acoust.Soc.Am.Suppl.181,S78 ( 1987); J. N. Larar, J. Schroeter,and M. M. Sondhi, J. Acoust. Soc. Am. 1:15 Y3. ARMA Murray Hill, NJ 07974) J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 Suppl. I 82, S54 (1987)], the codebookentrieshave to be modifiedto accommodate differentnominalvocal-tractsizes.Two differentwaysof speakeradaptationare compared.In the first approach,the codebookis comprisedof severalsubcodebooks, eachfor a differentsizeof the vocal tract.In thiscase,a methodfor determiningtheoptimalsubcodebook for eachspeakeris needed.Alternatively,in the secondapproach,the LPCrepresentation of the originalspeechis scaledfor accessto a constant tract-sizecodebook.Then, the codebookentry foundis inverselyscaled for synthesis. Here,a methodfor determining theoptimalscalefactoris required. 114th Meeting:AcousticalSocietyof America S54 2:OO Y6. Speechformanttrajectoryestimationusingdynamicprogramming with modulated transition costs. David Talkin (AT&T Bell Laboratories, Room2D-410, 600 MountainAvenue,Murray Hill, NJ 07974) A newalgorithmto trackautomatically speech formantfrequencies havebeendeveloped. Dynamicprogramming isusedtooptimizeformant trajectory estimates by imposing appropriate frequency continuity constraints.The continuityconstraints aremodulated by a stationarity function.The formantfrequencies areselected fromcandidates proposed by solving fortherootsofthelinearpredictor polynomial computed periodicallyfromthespeech waveform.The localcostsof all possible mappings of thecomplexrootsto formantfrequencies arecomputed at eachframe basedon thefrequencies andbandwidths of thecomponent formantsfor eachmapping.Thecostofconnecting eachof thesemappings witheachof themappings in the previous frameis thenminimizedusinga modified Viterbialgorithm.Two sentences spokenby 88 malesand43 femaleswere analyzed.The first three formantswere trackedcorrectlyin all sonorant regionsin over80% of thesentences. Theseperformance resultsarebased onspectrographic analysis andinformallisteningto formant-synthesized speech. 2:15 Y7. Somepropertiesof autoregressive modelrelatedcepstrum.BiingHwang Juang and David Mansour (AT&T Bell Laboratories,600 Mountain Avenue,Murry Hill, NJ 07974) Autoregressive modelrelatedcepstrum,or, briefly, LPC cepstrum, recentlygainedrenewedattentionbecause it wasshownto bean effective representation in speechrecognizerdesigns[B. H. Juang,J. G. Wilpon, and L. R. Rabiner, Proc. IEEE ICASSP-86, 765-768 ( 1986); B. Hanson and H. Wakita, Proc. IEEE ICASSP-86, 757-760 (1986); Y. Tohkura, Proc. IEEE, ICASSP-86, 761-764 (1986) ]. In this paper,someproper- tiesoftheLPC cepstrum areinvestigated thatareof importantconsideration whenthe recognizeris to be deployedin situationswheremismatch betweenthe trainingconditionsand the testingconditionsmay potentially occur.More specifically, the effectsof LPC analysisorder, additive noise,and pole movementsupon somekey characteristicsof the LPC cepstrumareconsidered, suchastheaverageandthe normof thecepstral vector.A simplerelationshipamongthe cepstralcoefficients, the predictor coefficients,and the reflection coefficientsis established,and unlike the Itakura-Saitodistortions,it is shownthat the L 2 distortionbetween twocepstraobtainedfromdifferent-order LPC modelsof thesamespeech data may containan inherentnonzerobias.Also consideredare the effects of additive noise,usingan additive power spectrummodel and a constrainedARMA model (sum of an all-polemodeland a constant).It is shownthat the norm of the LPC-cepstralvectorshrinksas a resultsof noisecontamination,comparedto that of the cleanspeech.Further, moving the polesof an LPC-modelspectrumis shownto be equivalentto multiplyinga powerseriesto the cepstrumand thusLPC-modelbandwidth broadeningand poleenhancement can be easilyaccomplished in bridgingtheencryptedsignalsin a singleB-voicechannel)isdiscussed. In particular,an enhancement procedureusinga prefilter,"dither" noise andan adaptivepostfilteris proposed. The systemis basedon an all-pole modelfor thedegraded speech. Thecoefficients of thepostfilterareeasily derivablefromtheall-polefilterparameters. Informallisteningtestshave shownthat theenhanced speech isnear"toll" quality. 2:45 Y9. Invariant acousticcuesin stop consonants:A cross-languagestudy usingthe Wignet distribution.H. Garudadri (Electrical Engineering, University of British Columbia, Vancouver,British ColumbiaV6T 1W5, Canada), J. H. V. Gilbert, A-P. Benguerel(Audiology and Speech Sciences,Universityof British Columbia,Vancouver,British Columbia V6T IW5, Canada), and M.P. Beddoes (Electrical Engineering, Universityof BritishColumbia,Vancouver,BritishColumbiaV6T 1W5, Canada) Acousticinvariancewasinvestigated in 200 uoaspiratedstopsfrom English,Telugu,andFrench,usingtheWignerdistribution. Thebilabials were characterizedeither by absenceof any noticeableburst,or by an energyconcentration in thelow-frequency regionduringtheburst,when present.The burstof alveolarswas mostlydiffusein nature,but sometimesbecamecompact,especially whenit wasfollowedby a backvowel. Theburstofvelarshada compactspectrum centered nearF2ofthevowel, whenfollowedbybackvowels, butwasmostoftendiffusewhenfollowed by a front vowel.The burstof all dentalfrom Teluguwasfoundto be diffuse.Whentheburstof alveolarswascompact,or thatof velarsdiffuse, the durationfrom the startof the burstto the startof the voweland F2 of the vowelwerefoundto be reliablecuesfor placeof articulation.The diffusenatureof bilabialsand alveolars,reportedby Lahiri et al. [J. Acoust.Soc.Am. 76, 391-404 (1984) ] and by the authorsreferenced therein,canbe partlyattributedto the temporalaveraging of the burst spectrum andthediffusevowelspectrum, whichisinevitable in Fouriertypeanalyses[Garudadriet al., J. Acoust.Soc.Am. Suppl.I 79, S94 (1986) ]. Unlike in otherinvestigations, no apparentpatterncouldbe foundfor the tokenswronglyclassified.[Work supported,in part, by NSERC, Canada. ] 3:00 Y10. A new mathematical model for intonation: Historical motivation and application. Deborah Rekart (Department of Speech Communication,The PennsylvaniaState University, University Park, PA 16802), and David Drumright (Department of Mechanical Engineering,The PennsylvaniaStateUniversity,UniversityPark, PA 16802) Early attemptsto characterizeEnglishintonationwere theoretical descriptions in termsof tones,tunes,andsegmental pitchlevels.Recent acousticdescriptions of intonationhaveinvolvedthe perceptualdeterminationof F0 contoursin termsof globaldeclinationlinesor asbreath the cepstraldomain. Consequently, evaluationof some frequency- groupswith terminaland nonterminalslopes.Liebermanet al. (1985) weighteddistortionmeasuresthat are believedto be desirableundercertain conditionsbecomes straightforward via cepstralprocessing. showedthat F0 contourscouldbedescribedobjectivelyby all-pointslinear regression lines,but thisdid not yielda compactcharacterization of the contours.Hirst (1983) developed a modelbasedon nonlinear(parabolic) function,but it is not unifiedin that it resetsparametersfor syllablesindependently. Attemptsto compareintonationcontoursin different languagesand dialectshavebeenbasedon the existingmodels.In the followingabstracta mathematicalmodelcapableof compactlyquantifying intonationcontoursusingfour parameters is proposed[Drumright and Rekart (1987)]. This paper demonstrates how the modelcan be appliedto analyzeforeignlanguage, non-native, anddialectaldifferences in the intonationof statementsand questionsin spontaneous and read speech. 2:30 Y8. Speech enhancement in crypto-bridge communication systems. Daniel Lin and Yacov Yacobi (Bell Communications Research, 435 South Street, Morristown, NJ 07960) In secureaudioteleconferencing, a crypto-bridgesystemsynchronouslyaddssimultaneous encryptedsignals,moduloa knownnumberP, andbroadcasts theresultto theparticipants. Eachterminal,usinga secret key, then decryptsthe modular sum of encryptedsignalsto obtain the desiredordinary sum of plain text signals.Suchsecrecysystemsare prov- ablysecure,assuming the existence of one-wayfunctions(e.g.,DES) for the modularsum cipher [E. F. Brickell et al., CRYPTO-87]. In this paper,the implementation of additivecryprosystemsin a narrow-band (2B + D) ISDN environmentare considered.The problemof noisereduction for 6-bit linear PCM speechcoding (an essentialconstraintfor $55 d.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 3:15 Yll. A new mathematical model for speech intonation: Theory and hardware/softwarerealization. David G. Drumright (Department of Mechanical Engineering,The PennsylvaniaState University, University 114thMeeting: Acoustical Society ofAmerica S55 rationis controlledby four overallparameters. The modelhasbeenrealizedandtestedin softwareformby a programthat readsan actualintonation curve [producedby the programPIG and describedin J. Acoust. Soc.Am. Suppl. 1 gl, S78 (1987)] and setsthe four parametersby an iterativematchingprocess.The hardwarerealization,by a circuit involving coupledoscillators,invitescomparisonwith analogous neuralstruc- Park, PA 16802), and DeborahM. Rekart (Departmentof Speech Communication,The PennsylvaniaState University,UniversityPark, PA 16802) A mathematicalmodelbasedoncombinations of sinusolds andexponentialdecayis shownto be superiorin representing actualintonation contours.The modelis generative ratherthanad hoc,in that theconfigu- WEDNESDAY AFTERNOON, 18 NOVEMBER tures. 1987 TUTTLE CENTER ROOM, 1:00 TO 2:50 P.M. SessionZ. Physical AcousticsIV: Recent Advancesin Fiber Optic Sensors David L. Gardner, Chairman NationalOceanic andAtmospheric Administration, CodeN/CGx3,Rockville, Maryland20852 Chairman's Introduction--l:0fi Invited Papers 1:05 Zl. Acousticsensordevelopmentat NRL. A. Dandridge (Code 6574, Naval ResearchLaboratory, Washington,DC 20375) The development of fiberopticacousticsensors at the Naval ResearchLaboratorywill be reviewed.The talkwilldescribe thefundamentals ofthedesign andoperation offiberopticsensors. Themajorareaofthistalk willbethetransduction mechanism ofacoustic energy toopticalphase shift.Methods todetectthephase shifts andtoprovide a linearvoltage outputwillonlybebrieflydiscussed. Thefrequency regimes tobecovered range from a few tensof Hz to MHz. In the low-frequency regime(up to severalkHz), wherethe pressureis hydrostatic,theroleof thematerialraGdullof fibercoatingsandbulkmandrelswill beconsidered. A numberof sensor designs for operation in thehigher-frequency regimes, whereinertialeffects do notpermitdirectaxial straintermsto play a major role, will alsobe described. 1:35 Z2. Multiplexingtechniques for interferometriefiber-opticsensors.A.D. Kerseyand A. Dandrige(Code 6574, Naval ResearchLaboratory,Washington,DC 20375) Considerable research interestispresently beingdirectedtowardthedevelopment of all-fiber,multisensor networksfor usein arraysandapplications wherea largenumberof differentphysicalparameters are of interest, suchasthoseencountered intheindustrial process controlarea.A numberof"all-optical"approaches to themultiplexing of suchnetworkshavebeenproposed in recentyears.Thesetechniques aremostattractive for usein a wide rangeof applicationareas,asthe sensorsoperatepassivelyand can be remotely locatedfrom thesourceanddetectionelectronics. Furthermore,theoverallsystemcomprising the sensors andfiberlinks retainsa low susceptibility to electromagnetic interference. The principleof operationof multiplexingtechniques basedontime,frequency, andcoherence-division addressing schemes will bedescribed. Thepaperwill compareexperimentally reportedperformance featuressuchasthe multiplexingefficiency,sensitivity,crosstalk, etc.,and will discuss practicallimitationsof the variousmethods. 2:05 Z3. Thermalfioiselimitationsin u fiberopticselsmiesensor.S. L. Garrett,T. Hofler(PhysicsDepartment, Code 61 Gx, Naval PostgraduateSchool,Monterey,CA 93943), and D. L. Gardner (National Oceanicand AtmosphericAdministration, Code N/COx3, Rockville, MD 20852) Opticalfibersareneithersensitive norselective asa transduction medium.The highsensitivityof mostfiber opticsensors hasbeena consequence of thefactthatopticalinterferometric demodulation iscapableof resolv- ingphase changes oforder1/zrad/x/-•.In afibersensor of10-mlength, thisisequivalent toaresolution ofone partin 10TM during a 1-sobservation using lightof900-nm vacuum wavelength. Forcomparison, agoodquality l-in. condenser microphone canresolvea relativechangein diaphram-to-backplate separation of onepart in 108duringa l-sobservation. Thisdetection advantage of 120dBhasobscured many"sins"inseveral fiberoptic sen,sor designs.This presentation will concentrateon the sensitivitylimitationsimposedby the fluctuation- S56 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S56 dissipation theorem [L. D. Landau andE. M. Lifshitz, Statistical Physics (Addison-Wesley, Reading, MA, 1969),2rided.,ChapXII ] at nonzerotemperatures andwill demonstrate that the thermallimitationsare entirely analogous tofluctuations observed in "mirrored galvanometers" thatalsouse"optical leverage" to increase sensitivity. Theresults willbeapplied totheperformance ofa seismic sensor witha sensitivity thatis limitedbyitsintrinsic thermalnoise.[Worksupported bytheOfficeof NavalTechnology andtheOfficeof Naval Research.] ContributedPaper 2:35 ZA. A compactfiber optic laser probe to monitor surface vibration displacements. Jacek Jarzynski,Dewon Lee, and Yves H. Berthelot (Schoolof MechanicalEngineering, GeorgiaInstituteof Technology, Atlanta, GA 30332) An heterodyne laserinterferometer isusedtomonitorthephasemodulationofthelaserlightscattered froma vibratingsurface. A laserbeamis splitintotwoarmsof an interferometer wherelightis guidedthrough singlemodeopticalfiberssoasto preserve phaseinformation. Therefer- WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 enceand frequency-shifted signalsare recombinedin a multimodefiber tightlyconnected to a receivingphotomultiplier. Miniaturegradientrefractiveindex (GRIN) lensesare usedto providemaximumefficiency andto maketheprobeascompactaspossible. Bothnormalandtangential components of thesurfacedisplacement canbedetermined witha probe illuminatingthe vibratingsurfaceat normalincidenceand at two mutually perpendicular off-normaldirections.The tangentialcomponentof the surfacedisplacement is foundfromspecklereflection of thenonperfectly flat vibratingsurface.This compactprobecanbe positioned at somedistance( 10-50 cm) from the vibratingsurfacefor truly nonintrusive measurements. PREFUNCTION AREA, 1:00TO 3:00 P.M. SessionAA. Psychological and PhysiologicalAcousticsIV (PosterSession) Craig Formby, Chairman Departmentof Communicative Disorders/Neurology, J. H. Miller Health Center,Universityof Florida,Box J-174, Gainesville, Florida 32610 ContributedPapers All posters will bedisplayed from 1:013 to 3:00p.m.To allowcontributors theopportunityto seeotherposters, contributors of odd-numbered paperswill beat theirposters from 1:00to 2:00p.m.andcontributors of evennumberedpaperswill beat theirpostersfrom 2:00to 3:00p.m. AA1. Diseriminability,loudness,and maskingin the rat: A confirmation and extension. Thomas G. Raslear (Department of Medical Neurosciences, Walter Reed Army Instituteof Research,Washington, DC 20307-5100) Rats discriminatedbetweentwo sound-pressure levels(SPL) of a puretone:standard(STD) SPLsof 84 and74 dB, andcomparison(CO) SPLs4, 14,and24dBbelowSTD weretestedin quietand60-dBnoiseat 4 and 12.5 kHz (24 c0nditions). The decibel differencebetween STD and CO accountedfor only 43.52% of the variancein the signaldetection measureof sensitivity(d') acrossconditions,whereasthe loudnessdiffer- ence(LD = STDø3• - COø'3s) accounted for 89.82% ofthe variancein d '. Theseresults confirmandextendprevious observations that:( 1) equal decibeldifferences arenot equallydiscriminable [R. Piertel,J. G. Sherman,S.Blue,andF. Hegge,J. Exp. Anal. Behav.13, 17-35 (1970) ]; (2) loudness for therat increases asa powerfunctionof SPL, with an expo- nentof0.35 JR.Pierrel-Sorrentino andT. G. Raslear, J.Comp.Physiol. Psychol.94, 757-766 (1980) ]; ( 3 ) maskedloudnessisa linearfunctionof loudness in quiet [T. G. Raslear,R. Pierrel-Sorrentino, and F. Rudnick, Behar. Neurosci. 97, 392-398 (1983) ]. L. Brill (Chicago ZoologicalSociety,BrookfieldZoo, Brookfield,IL 60513andParrelyHearingInstitute,LoyolaUniversityof Chicago,6525 N. SheridanRoad, Chicago,IL 60626), Martha L. Sevenich,Timothy J. Sullivan,JanetD. Sustman,andRonaldE. Witt (ChicagoZoological Society,BrookfieldZoo, Brookfield,IL 60513) A dolphin (Tursiopstruncatus)was conditionedto performa discriminationtask,by meansof echolocation, in a "Go/No-go" paradigm whilestationedin an underwaterhoopand wearingsuctioncupsoverits eyes.To investigatethe hypothesisthat the fat-filledlowerjaw of odontocetecetaceansprovidesan acousticalpathwayto the inner ear, the dolphinwasadditionally requiredto performthetaskwhilewearingeitherof two rubberhoodsdesigned to coverits lowerjaw. Onehood,constructed from nonfoamedneoprene,allowedreturningacousticsignalsto pass, whiletheotherhood,constructed fromclosed-cell neoprene,substantially attenuatedsuchsignals.The dolphin'sperformance wassignificantly hindered whilewearing theattenuating hood(p < 0.001,X 2) asopposed to its wearingthe hoodmadeof nonfoamedneopreneor no hoodat all. Acousticaldata tape recordedduring the experimentindicatethat, with few exceptions, the spectralpeaksof the dolphin'soutgoingecholocation signalsaveragedbetween30 and 50 kHz. Theseresultssupportthe hypothesisand agreewith previouselectrophysiological data that suggest AA2. Evidence for an acoustical pathway to the inner ear through the that thelowerjaw is involvedin the transmission of high-frequency sig- lower jaw for an echolocatingdolphin(Tursiopstruncatus).Randall nals to the inner ear. S57 d.Acoust.Sec.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S57 AA3. Modulationand gap detectionwith filtered and broadbandnoise signals.C. Formby (Departmentsof CommunicativeDisordersand pieallylimitedextentsat differentrates.Results appearconsistent with augmentinginfieldinhibition. Neurology,Universityof Florida,Gainesville,FL 32610), K. Muir, J. Uranga,C. Lynn,andA. Ahlers(Departmentof Speech, University of Florida, Gainesville, FL 32610) The higheraudiometric frequencies arguablyarethemostimportant for sometasksrequiringfinetemporaldiscriminations. Bothmodulation andgapdetection thresholds improve withincreasing centerfrequency of noisebands.For modulationdetection,however,not all investigators are convinced that improvedperformance isnecessarily dependent on sensitivity at the highercenterfrequencies. They pointout that the high-fre- AA6. The effectsof pedestallevel on incrementalauditory responses in humans: Loudness matching study. Hiroshi Riquimaroux and Peter Dallos (Auditory Physiology Laboratory, Northwestern University,2299SheridanRoad,Evanston,IL 60201) The effectsof pedestallevel (L B) on incrementalauditoryresponses wereinvestigated in normal-hearing humansubjects withamplitudemod- quency advantage formodulation detection maybea consequence ofcon- ulated (AM) white noise by meansof a loudnessmatching procedure. comitantincreasesin stimulusbandwidthas a functionof increasing centerfrequency of thenoiseband.To determine whethersignalfrequen- The methodof adjustmentwasusedto findequallyloudincrements(AL) for variousstandardsALs andL Bs ( 30-80 dB SPL). The L B for standard cyor signalbandwidth ismoreimportantfor modulation andgapdetection, modulationand gapdetectionthresholds weremeasuredfor lowpass (LP) filterednoise (fc--4000 Hz, bandwidth=4000Hz), high-pass (HP) filterednoise(fc = 4000Hz, bandwidth = 2500Hz), loudness of AL isindependent ofL B.The resultsshowthatloudness of •L and broadband(BB) noise (bandwidth= 6500 Hz). Modulation and gapdetection thresholds fortheBBandHP signals weresimilar,andwere betterthan thresholdsfor the LP signal.The ability to hear the higher frequencies of the noisesignalapparentlyis moreimportantfor these tasksthan is perceptionof signalbandwidth.[Researchsupportedby Divisionof Sponsored Research,Universityof Florida,and American •L waskeptconstant at 30dBSPL.Thehypothesis tobetestedwasthat appears to bealmostindependent ofL• for verysmallAL ( < 6 dB). The findingsalsoindicatethe possibilitythat the stimuluspeaklevel (Lv) mightbe involvedin the determination of loudness of AL whenAL is large.The datademonstrate that the slopeof loudness growthfunction increases asL B increases. For lowL Bs,theslopehasa tendencyto asymptote to the referenceline with the slopeof one. The resultsagreewith earlierJND data. It is concludedthat the hypothesisappearsto be supportedby the presentdata for verysmallALs. Cancer Society.] AA4. Effect of click rate and delay on breakdown of the precedence effect. Rachel K. Clifton (Department of Psychology,University of Massachusetts,Amherst, MA 01003) and Richard L. Freyman (DepartmentofCommunication Disorders,Universityof Massachusetts, Amherst, MA 01003) When clicksare presentedfrom two loudspeakers, with one output leadingthe otherby a few ms,a listenerlocalizesthe soundsolelyat the leadingloudspeaker,a phenomenonknown as the precedenceeffect. However,the precedence effectcanbe disruptedfor severalsecondsif the leadingand lagging(echo) locationsare quickly switched[Clifton, unpublishedmanuscript].Click trainswerepresented from two loudspeakers, placed90øright and left off midlinein an anechoicchamber,with laggingclickdelaysrangingfrom2-9 ms.Clickrateswere1,2, 4, 6, and8/ s. Under someconditions,subjectsreported hearing clicks from both loudspeakersimmediatelyfollowing the switch in leading and lagging outputs,indicatinga breakdownof the precedence effect.The lengthof timebothclickswereheardfollowingthe switchwasaffectedby bothecho delay and click rate. Inhibition of the precedenceeffectwas strongest whenrate of clickswasslowerand delayswerelonger,that is,just under the subject'sechothreshold.At shorterdelaysand fasterratesthe precedenceeffectwasexperienced throughoutthe trial. AA7. The effectsof pedestallevel on incrementalauditory responsesin humans:Brainstemresponse(ABR) study.Hiroshi Riquimarouxand Peter Dallos (Auditory Physiology Laboratory, Northwestern University,2299 SheridanRoad,Evanston,IL 60201) The effectsof pedestal level(L•) on incremental auditoryresponses wereexaminedin normalhearinghumanadultswith amplitudemodula- ted (AM) whitenoiseby usingtheauditorybrainstemresponse (ABR). The waveV latencyof ABR, yieldedby a noiseincrement(AL) with various Lss ( 30-80dBSPL), wasmeasured. Thehypothesis ofthisstudy is that the waveV latencyderivedby AL is invariantwith L B. The results showthatthewaveV latencyappearsto bealmostindependent of L• for verysmallAL ( <6 dB). Thedataarecompared withthoseofthecompanionloudness matchingstudy.Similartendencies areseenin tbecharacteristicsof iso-waveV latencycontoursand equalloudnesscontoursfor relativelysmallALsor for lowLss. The growthfunctionsobtainedfrom the two experiments, however,are different.In the loudness matching studythe slopeof loudness growthfunctionsincreases asLs increases, whilein the ABR studythe slopeof growthfunctionsof waveV latency increasesonly for low L s. AAS. Testing children and the aged with computerized audiomerry. AAS. Inhibitory mechanismsand cortical processing.D. M. Daly, D. D. Daly (Box 210855, Dallas, TX 75211), J. W. Drane (Auburn University,Auburn, AL 36849), and J. A. Wada (University of British Colombia, Vancouver, British Columbia V6T 1W5, Canada) When balanceof inhibitionsdeclines/fails,patientswith seizuresin auditory cortex report certain synthetic [be]-[de]-[gc] as nasals, "bleats,"and then undifferentiatedbuzzes.Effectsof increasedinhibition canappearwithauditorycortexin volumesurrounding disinhibited areas (evencontralaterally):Patientsreportsamerangeof stimulias clearly identical[J. Neurophysiol.44, 200-222 (1980)]. Here a casewith seizuresin visualcortexand post-ictallyimpairedauditorycomprehension (but normalperipheralacuity),isreported.Testingincludedprerecorded setsof [be]-[de]-[gc], [ge]-[ye], and [be]-[we] presentedthrough headphones monaurallyandbinaurally.Duringepisodes with morefrequentseizures,patienthad intervals(somelastingnearlyan hour) when [de] boundariesfor [be]-[de] and [de]-[ge] increasedapproximately 200 Hz; [be]-[wc] and [gc]-[ye] boundaries increased 40-50 ms (with stimulilessthan 50 msnowreportedasvocalic).The BDG stimulicover tonotopicallydifferingextentsof cortex;BW or GY stimulicovertonoto- S58 J. Acoust. Soc.Am.Suppl.1, Vol.82, Fall1987 Michel Pieard (University of Montreal, Montreal, QuebecH3C 3J7, Canada), Henry J. Ilecki, and Sames D. Baxter (Royal Victoria Hospital/McGill University,Montreal,QuebecH3A 1A1, Canada) Usingcomputerized audiometryin a groupof noise-exposed workers has beenshownto be a reliableand valid meansof assessing hearing sensitivity[ M. Picardetal., Proceedings: Computer /lpplications to Canadian Health, pp. 127-130 (1987) ]. Basedon this study,an investigation wasconducted to similarlydeterminethefeasibilityof implementing likemethodology in twoclinicalpopulations typicallyregardedas"difficultto test" (viz., geriatricand pediatricgroups).Childrenwereaged8 to 12; seniors,65-81 years.Reliability was assessed by relatingair- and boneconductionthresholds wheretherewasno evidenceof middle-earpathology. Validity was measuredby comparingair-conductionpure-tone thresholds(0.5, 1, 2, 3, 4, and 6 kHz) obtained under two test modalities, computerizedaudiometry and conventionaltesting performedby a trainedexaminer.Bothprocedures followedANSI S3.21-1978.Findings in both agegroupswerecomparableto thosereportedin the previous study.Coefficients of reliabilityremainedequallyhighacrossfrequencies regardless of testingmethod.As well,highcorrelations betweenconventional and computerized audiometrywerefound.Complementing this 114thMeeting: Acoustical SocietyofAmerica S58 observation, a PCAsupported thecriterion-related validity ofcomputerizedaudiomerry. [Worksupported by MRC. ] AA9. Frequency specificity of theeochlearactionpotentialvarieswith stimulus presentation rate. Alex N. Salt, Arti R. Vora, and Ruediger Thalmann(Department of Otolaryngology, Washington demonstrate the presence in the OC of specific subtypes of certainproteins.In thecaseofcontraction associated proteins, forinstance, thepresenceof nonmuscle actin (betaand gamma),threetypesof nonmuscle tropomyosin (Tm:4,Tin:5,andTin:8) andnonmuscle lacticdehydrogenase(LDH B) wasdocumented. Anotherimportantroleof 2D-PAGE is thedemonstration of previously unknownproteins. The presence in the OCoftwohighlyprominent proteins oflowmolecular weightandstrongly acidicpH(assumed to bespecific to theOC) wasshown.Accordingly, theyweretermedOCP-I andOCP-II [I. Thaimannetal., Arch. Otorhin- University,SaintLouis,MO 63110) olaryngol.226, 123-128 (1980)]. It was found that small amountsof It iswelldocumented thattheamplitude of anaveraged cochlcar actionpotentialdecreases astherateof stimulus presentation isincreased. In thepresent study,frequency regions contributing totheAPelicited by toneburststimuliat ratesfrom4 to80/shavebeeninvestigated in guinea pigsusing a tonemasking procedure. Derived responses corresponding to specific regions wereobtained by subtracting AP responses in thepres- OCP-I andOCP-II arepresentin spiralligamentandspirallimbusand moresubstantial amountsin basilarmembrane. Neitherproteinisdetectablein striavascularis. Likewise,noproteinsresembling OCP-I or OCP- enceof a low-levelcontinuous maskingtonefromthosein theabsence of masker.Thisisa modification of thetechnique described by Hoodet al. [J. Acoust.Soc.Am. Suppl.1 81, S8 (1987)]. Whenstimulationrateis II were found in tissues other than inner ear. While some biochemical featuresof the two proteinshave in the meantimebeencharacterized, their functionalsignificance remainsobscure.[Work supportedby NIH. 1 increased, thederived response amplitudes decreased neartheprobefrequencyregion,whileamplitudes at outlyingfrequencies werelessmarkedlychanged. Thesedatasuggest thecontribution to theAP fromthe region corresponding totheprobefrequency becomes lessathighstimula- tionrates.Withan80/sstimulation ratethecontribution fromthisregion isnegligible. Thedegradation offrequency specificity ofresponses ispre- AAll. Ultrastructural correlatesof hair cell motility. B. N. Evans and A. Yonovitz (GraduateSchoolof BiomedicalSciences, The University of Texas Health ScienceCenter at Houston, Houston,TX 77030) sumedto arisefromeachstimulusactingasa "forwardmasker"for the following stimulus. To optimize thefrequency specificity ofAP responses Nonstimulatedas well as outer hair cellsshowingmechanicaldis- to toneburststimuli,the rate of presentation mustthereforebe considered. [Work supportedby NIH.] placementsto electricaland chemicalstimulationwere examinedultrastrueturally to assessthe lability of a rhodamine-labclcdactin structure AA10. Proteins of the organ of Corti (OC}. IsoldeThalmann, GertraudThailinger, and RuedigerThaimann (Department of Otolaryngology, Washington University,SaintLouis,MO 63110) extendingbetweenthe cuticularplateand the peri-nuclearregion.Hair cellsof varyinglengths(20-80 pro) werestimulatedand immediately fixedfollowingmechanical displacement. All cellsrevealeda distinctnonmembraneboundcomplexin crosssection,but were infrequentlyobservedin longitudinalsections.The cross-sectional area of the structure increased in successive ascending sections. This structurespiralsapieally in a nondirectmanner,buteventuallyconsolidates intoa primarymassto becomecontiguouswith the inner leafletof the cellsplasmalemma.A localizedmicrotubulenetworkwasfoundto passmostprominantlyin the thin basalaspectof the complex.The cross-sectional orientationappears to optimizethe visualizationof this potentiallyimportantaspectof the outer hair cellsanatomy. The majorityof proteinsfoundto datein OC wereidentifiedby immunohistological methods. Whilethesetechniques arepowerfultoolsfor localization ofproteins, theyfrequently lacksatisfactory specificity. Twodimensionalpolyacrylamidc gel electrophoresis (2D-PAGE) is able to WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 UM LEARNING CENTER, 1:00TO 2:15 P.M. Session BB. Psychological andPhysiological Acousti csV: Frequency Discrimination andPitchPerception William M. Hartmann, Chairman Physics Department,MichiganState Unioersity, EastLansing,Michigan48824 ContributedPapers 1:00 BBl. Tests of a modified energy-detector model of frequency discrimination. David $. Emmerich, Wolfgang Ellermeier, and Brenda Butensky{ Departmentof Psychology,StateUniversityof New York at StonyBrook, StonyBrook, NY 11794) Henning[J. Acoust.Soc.Am. 42, 1325-1334(1967)] proposed a modifiedenergy-detector modelof auditorydiscrimination anddetection. Thismodel,aswellasseveralothermodels,predictsthat randomvaria- tionsof theamplitudes of thesignals to bediscriminated in a frequencydiscrimination experimentshouldmakethetaskmuchharder.However, $59 J. Acoust.$oc.Am.Suppl.1, Vol.82, Fall1987 Henning[J. Acoust.Soc.Am. 39, 336-339 (1966)] reportedthat, for signalfrequencies lessthan4000Hz, variationsin amplitudedid not lead to theexpectedimpairmenton the frequency-discrimination task.A reexaminationof this questionindicatesthat when the appropriatecontrol comparisons aremade,subjects doin factshowthepredicted impairment. Neverthelessa secondexperimentsuggests that this resultmay not providea goodtestof Henning's(1967) energydetectionmodelsincevariationsin pitch with intensity,which are not consideredin the model,are likelyto beinvolved.For thisreason,a thirdexperimentwasconducted in orderto providean independenttestof the model.The resultsof this third experiment do not supporttheenergy-detector model. 114thMeeting:Acoustical Societyof America S$9 1:15 Previous research hasdemonstrated theinfant'sabilityto extractthe virtualpitchofharmonic tonalcomplexes. In thepresent study,weuseda conditioning procedure toassess 7-month-old infants'perception ofpiteh for inharmonic tonalcomplexes. Infantsinitiallydiscriminated a pitch change from 160to 200Hz or viceversain harmonic tonalcomplexes, BB2. Detectingdynamicfrequencyshiftsof simpleand complextones. RobertP. Carlyonand Richard$. Stubbs(MRC Instituteof Hearing Research, UniversityPark,NottinghamNG7 2RD, UnitedKingdom) which containedeight consecutive harmoniesbut no fundamentalfre- Thresholds for thedetection offrequency modulation bya singlecycle ofa 12-Hzsinusoid wereobtainedfor carriersthatweresimpleor complex tones.Two phases of modulatorwereused,producingpercepts of either upwardor downwardfrequencyglidesduringthemiddleof thestimulus. To ensurethatsubjects baseddetection on frequency mooement, boththe phaseof the modulatingsinusoidand the startingfrequencies were randomizedfrom presentation to presentation. For pure-tonecarriers, thresholds ( = minimumdetectable frequency swing/carrierfrequency) wereconstantfor carrierfrequencies of 800-2400Hz andincreased below 800Hz. Thresholds werelowerfor modulations producing theperceptof an upwardglidethan for "downward"glides.This differencewasmuch greaterat low cartierfrequencies. Whenthecarrierwasa three-componentharmoniccomplex,thresholds werealmostindependent of itscenter frequencybut increased with decreasing fundamentalfrequency.Thresholdsfor harmonicand for inharmoniccomplexes werecompared.These resultshaveimplicationsfor the recordingof voicepitch {e.g.,in signalprocessing hearingaids). quency. Afterinfantsdiscriminated theharmonic complexes, eachpartial'sfrequencywasshiftedupwardby 30 Hz to renderthe soundsinhar- tannic. All infantswho discriminated the inharmoniccomplexes proceeded to a categorization stagein whicha randomized sequence of threeinharmonic complexes (containing six partials)comprised each pitchcategory sothatthespectral characteristics of thetokenschanged continuously whilethepitchcategory changed onlyonchange trials.The harmoniccomplexes werediscriminated by 19 of 22 infants,while 16 infantssuccessfully discriminated theinharmonic sounds. However, only 11infantscategorized spectrally different sounds consistently withtheir pitches, suggesting thatinfants hadmoredifficulty extracting pitchfrom inharmonic complexes thanfromharmonic onespresented in previous research.[Work supported by NSF andNICHD.] 2:00 1:30 BB3. Central spectrum models of dichotic edge pitches. W. M. BBS. Pitch and timbre of inharmonictone complexes of Baiinese Hartmann (PhysicsDepartment,Michigan State University,East "GenderWayang"instruments. WernerA. Deutseh(Department of Lansing,MI 48824) Sound,AustrianAcademyof Science, A-1010,Vienna,Austria)and Franz Fodermayr (Department ofMusicology, University ofVienna,A- Thecentralspectrum modelof Raatgever andBilsen[J. Aconst.Soc. 1010 Vienna, Austria) Am. 80, 429-441 (1986) ] successfully accounts for a numberofdichotic- noisepitcheffects andfor thelateralization of thepitches. The modelis, however, physiologically implausible. An improved model,based upona neuraldelaylinehavingcoincidence cellsthatobeythestatistical rulesof Thetonesof thescaleoftwoBalinese "GenderWayang"instruments havebeenstoredon computerdiskand are usedas testtonesin pitch detectionexperiments. The subjects wereinstructedto identifythepitch of theinharmonicinstrumenttonesby tuninga sinusoidal testtoneon the appropriate pitchheightin a computer controlled up/downadaptiveprocedure.The experimental resultssupportthegeneralambiguityof pitch perception in caseofinharmonictonecomplexes, thougha certainpreferenceof thesubjects on thelowestpartialcouldbeobserved. Thisperceptual ambiguitycouldbe increased furtherwhenthe firstpartialof the instrumenttoneswasremovedby meansof narrow-band digitalfiltering. Basedonfrequency analysis dataof thenaturalsounds, a synthesis of the figuration"melody"of theGenderWayangpiece"Lagndelem" hasbeen performed in whichtheinharmonic complex toneshavebeenreplaced by harmonic ones.Thecomparison ofall three,theoriginalsoundprobes and the filteredand the synthesized tonesshowsthe strongdependence of pitch on timbre.The resultsof the pitch detectionexperimentsare discussed onthebasisofcriticalbandanalysis andthecurrenttheoryof pitch perceptionaspublishedby E. Terhardt (1972). StemandColburn[$. Aconst.SOc.Am. 64, 127-140(1978) ], retainsthe predictive poweroftheRaatgever-Bilsen model.It alsosuccessfully predictstheresultsof newexperiments wherethe Raatgever-Bilsen modei fails.Newexperiments measure therelativestrengths ofHuggins pitches HP + andHP --, thelateralization of thebinaural-edge pitch(BEP), thespectral densitydependence of theBEP,andthestrength of generalized BEP, wherethe interauralphaseshiftmay haveany magnitude. [Work supported by theNIH. ] 1:45 BB4. Infants' extractionof pitch from inharmonictonal complexes. MarshaG. ClarksonandRachelK. Clifton(Psychology Department, Universityof Massachusetts, Amherst, MA 01002) WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 UMS 4, 1:00 TO 3:05 P.M. SessionCC. UnderwaterAcousticsIV: WedgeModel Propagation Ding Lee, Chairman Naoal UndermaterSystemsCenter,New London, Connecticut06320 Chairman's Intruduetio•l:00 ContributedPapers CCI. Range-dependent cot-an frequenciesin a three-dlmensionai ocean environment. StewartA. L. QleggandSongR. Yoon (Departmentof OceanEngineering,FloridaAtlantic University,BocaRaton,FL 33431) Soundpropagation in a three-dimensional wedge-shaped oceanenvi- S60 J. Acoust.Soc.Am.Suppl.1, VoL82, Fall 1987 ronmentdemonstrates significantray curvaturein a directionparallelto thewedgeapex.The theoretical prediction of thisshowsthattheseeffects are moresignificantat low frequencies than at high frequencies; consequently,thethree-dimensional oceanactaasa range-dependent acoustic filter.This paperdescribes an experimental andtheoreticalinvestigation of pulsepropagation in a modelscalethree-dimensional wedge-shaped 114thMeeting:Acoustical SocietyofAmedca S60 ocean.In the experimental setupthe completeacousticfield couldbe measuredfor a givensoumepositionand the wedgeanglecouldbe changed. Theresults showgoodagreement withtheidealized modalsolutionof Buckingham, anddemonstrate howthe pulseis distortedas a functionof range.Theyalsodemonstrate howthecut-onfrequency for acoustic propagation increases asa functionofrangeparalleltotheshoreline. 1:20 CC2. Scatteringin hasinsor horizontally refracting modes:Which is it? C. H. Harrison (CAP Scientific,40-44 Coombe Road, New Malden, SurreyKT3 4QF, United Kingdom) In a largebasinthereare at leasttwo typesof 3-D ray pathsfrom a sourceto a distantreceivervia the slopingbasinedge(other than the directpath). Oneisthesimplescatteredrayat theseabed, andtheotheris the multiplereflectedray that steepens in shallowwaterbut finallyturns in the horizontalplanetowardsdeepwater (otherwiseinterpretedas a horizontallyrefractingverticalmodecontribution[C. H. Harrison,J. Acoust.Soc.Am. 65, 56-61 (1979)]). If the totalchangein headingof thelatterray is 20, itselevationanglewill neverexceed0 [C. H. Harrison, J. Acoust.Soc.Am. 62, 1382-1388(1977) ], andthepathmayexistdespitereflectionlosses in realisticenvironments. Therefore,if the topand bottomsurfaces areperfectlysmooth,therewill benoscattering andthere will certainlybemultiplereflection. Conversely, if thesurfaces arerough, themultiplereflection will beinhibitedbut therewill definitelybea scatteredreturn. Sincethereis scopefor misinterpretation of experimental results,this paperproposes someexperiments to distinguishthe two paths. America in Indianapolis,Indiana, 11-15 May 1987. The underwater acoustics communitywasinvitedto presentsolutionsto theseproblems andcomparethemagainsta numerically"exact"solutkmobtainedfrom theR. B. Evans'coupledmodemodel[F. B. Jensen, J. Acoust.Soc.Am. Suppl.I 81, S40(1987)]. Ourfiniteelementoceanacoustic propagation model[J.E. MurphyandS.A. Chin-Bing, J.Acoust. Soc.Am.Suppl.1 81,S9 ( 1987) 1to ProblemI isappliedhere.(ProblemI dealtwithshallow waterupslopepropagation in a wedge-shaped channel.)This finiteelementmodelgivesa full-wavesolution forrange-dependent acoustic propagationwith accuracythat rivalscoupledmodemodels.The model'scapability will be demonstrated by presentinga numericalsolutionto ProblemI that comparesfavorablyto the Evans'coupledmodemodel solution.[Work supportedby ONR/NORDA. ] 2:05 CC5. Soundinteractionwith a sloping,upward-refracting oceanbottom. HassanB.Ali, JuanI. Arvelo(NavalSurfaceWeaponsCenter, CodeU- 25, White Oak, Silver Spring,MD 20903-5000),Anton Nagl, and HerbertOberall(Department of Physics,CatholicUniversity, Washington,DC 20064) Results of a parobolic-equation (PE) codeareshownfortheproblem of soundinteractionwithan upsloping, upward-refracting oceanbottom. Theyarecomparedwith the resultsof a coupled-mode code[ J. F. Miller etal., J.Acoust.Soc.Am. 79,562 (1968)] fora range-dependent environment.Thisincludes bottompenetration effects at modecutoffanda study of howtheseareaffected bythesound-speed gradientandby theabsorptionin thesediment. The bottompenetration featuresarefurtherillumi- natedbyemploying Gaussian soundpulses andstudying theirpropagation,illustrating theeffects ofbottompenetration andreradiation backup into the water (as causedby the upward-refracting bottomgradient}, arrivingaheadof the water-bornepulse. 1:35 CC3. Diffraction of sound by hard strips and truncated wedges. Ivan Tolstoy(Knockvcnnic,CastleDouglas,SW ScotlandDG7 3PA) Startingfromrigorousdiffractionsolutions for single,hard,semi-infinitewedges, it haslongbeenpossible to construct approximate solutions for morecomplexangularsurfaces[ A.D. Pierce,J. Acoust.Soc.Am. 55, 941-955 (1974)]. Demonstrated hereisan exactprocedure thatformulatesthe multiplescatterbetweenverticesof an angularbodyvia the standardself-consistent algorithm.This allowsoneto obtainthecorrect diffractedfieldof eachvertex.Summingthesetogetherwith the relevant incidentandimagefieldsleadsto an exactsolutionfor thefull soundfield. Thisisdonehcrcfor thecaseof a planeharmonicincidentsoundfieldfor ( 1) a truncatedwedgeand (2) a hardstripof width l. Numericalresults for the latter modelare shownto comparesatisfactorily for all kl with experimentaldata [H. Mcdwin etal., J. Acoust.Soc.Am. 72, 1005-1013 (1982) ]. This theoreticalprocedureyields,in the planewavecase,a for- malestimateof theerrorincurredin the"doublediffraction"approximation (i.e., the procedurethat neglectsall but the first two termsof the multiple scatter series). Thiserrordecreases likeeikt/klforincreasing k/. It isthenpossible toconfirmtheoreticallytheobservation of Medwinetal. whoshowed that,forthegeometry oftheirstripexperiment, thisapproximationis actuallyverygoodfor kl>• 3 and quiteadequatein the region I < kl < 3, usingtheBiot-Tolstoyrigorousimpulsive pointsourcesolutionfor theperfectwedge[ M. A. BiotandI. Tolstoy,J. Acoust.Soc.Am. 29, 381-391(1957)] for their double-diffraction solution.[Work sup- 2:20 CC6. Sphericallysymmetricacousticpropagationacrossa fluid/fluid boundary. MichaelJ. Buckingham (Departmentof OceanEngineering, Room5-204,Massachusetts Instituteof Technology, Cambridge, MA 02139) The acousticfieldproducedby a pointsourceat thecenterof a fluid sphere,whichitselfisimmersedin a fluidmedium,isof interestin connection with ambient noise in the Arctic Ocean. A solution for the field in the internaland externalfluid domainsis presentedbasedon novelfinite Hankel transforms. The method involves internal and external Hankel transformations of the Helmholtzequation, whichintroduce explicitly theboundaryvaluesof thefieldandits spatialderivativeat thespherical interfacebetweenthe two fluids.By combining the resultantequations withtheboundary conditions (continuity ofpressure andnormalcomponentof velocity)all theunknown constants aredetermined, andon performingtheinverse Hankeltransforms, thefinalexpression forthefieldis obtained. The mathematical formalism of the finite Hankel transforms introduced herehastheadvantage of beingableto handlecomplicated boundaryconditions,whereaspreviousfinite Hankel transformtechniquesarelimitedto problems involvingDirichlet,Neumann,or mixed (i.e.,impedance) boundaries. [Thisworkwassupported by theOfficeof Naval Research,ContractNo. N00014-86-K-0325.] portedby ONR. ] 2:35 1:50 CCA.Anaccurate numerical solution toASABenchmark ProblemI using thefiniteelementmethod.Joseph E. Murphy(Departmentof Physics, University of New Orleans, New Orleans, LA 70148) and Stanley A. Chin-Bing(Naval Ocean Researchand DevelopmentActivity, NumericalModelingDivision,NSTL, MS 39529-5004} A SpecialSessionon Numerical Solutionsof Two BenchmarkProb- lemswaspresented at the 113thMeetingof the Acoustical Societyof $61 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 CC7. Sound transmissionexperimentsfrom an impulsivesourcenear rigid wedges.Saimu Li (ShandongCollegeof Oceanography, Quindao, Shandong, People's RepublicofChina)andC. S. Clay (Geophysical and Polar ResearchCenter,Universityof Wisconsin,Madison,WI 53706) Experimental sound transmissionsin air from a spark source to a small microphonewere made near rigid wedges.Two typesof experimentswere made. The first experimentswere transmissions within an approximately12øwedgeanda 52øwedge.The Biot-Tolstoywedgesolution [I. Tolstoy,WavePropagation (McGraw-Hill, New York, 1973)] was usedto calculatethe theoreticalimpulseresponse. The "free air" 114thMeeting:Acoustical Societyof America S61 transmission fromthe sparksourcewasconvolved with the theoretical transmission. The transmissions within thewedgegavefinitesetsof multi- plereflections or imagearrivalsanda diffraction arrival.Thediffraction wassensitive to leaksat thewedgeapex.Theoryandexperimentmatched. Thesecond setofexperiments wasmadeneara 270øwedge.Arrivalswere the direct reflectionand the diffraction.The image reflectionwhen the specular"reflectionpoint"wasverynearthewedgeapexwasof interest. Comparisons of dataandtheoretical signalsusingBiot-Tolstoytheory were excellent.Theoreticaldiffractionsignalscalculatedwith Trorey's theory had poor matches[A. W. Trnrey, Geophysics35, 762-784 (1970) ]. 2:50 CC8. Impulse response of density contrast wedge using normal coordinates.D. Chu and C. S. Clay (Geophysicaland Polar Research Center,Universityof Wisconsin,Madison,WI 53706) WEDNESDAY AFTERNOON, 18 NOVEMBER The exactimpulseresponse of a pointsourceto a rigid wedgewas derivedby Biot and Tolstoy [I. Tolstoy, WavePropagation(McGrawHill, New York, 1973)]. The solutionby the experiments hasbeenverified.The wedgehavingfree (pressurerelease)boundarieshasthe same formastherigidboundarysolution[W. A. Kinneyetal., L Acoust.Soc. Am. 73, 183-194 (1983) ]. Thesesolutionscannotbe applieddirectly to many actual casessuchas oceanfloor, wherethe ratio of densityof the basesedimentto that of the waterapproaches neitherinfinite(rigid) nor zero(pressure release). Thesolution istheimpulse response toanisovelocity (v• = v2)anddensitycontrast(Pl •P:) wedge.It is an extension of thesolutiongivenby Blot andTolstoy.The reflectionpart of the solution is an impulseseriesweightedby reflectioncoefficient with differentorder. The diffractionpart is the summationof the diffractionsdue to the individualimagesources. The amplitudesarea functionof thereflectioncoefficientand the numberof multiplereflections.[This work waspartly supportedby the People'sRepublicof China and the Officeof Naval Research.] UMS 5, 1:00 TO 3:05 P.M. 1987 SessionDD. Underwater Acoustics V: Noise Modeling Tsih C. Yang, Chairman Naval ResearchLaboratory,Code5123, Washington,DC 20375 Chairman's Introduction--l:00 ContributedPapers 1:05 DDI. Low-frequency noiseandbubbleplumeoscillations. W. M. Carey (Naval UnderwaterSystemsCenter,New London,CT 06320) and J. W. Fitzgerald(Consultant,New London,CT 06320) The evidencefor a wind dependencein low-frequencyambientnoise ( < 200 Hz) is sparsedue to the corruptinginfluencesof the radiated soundfromships.Measurements madewith hydrophones belowcritical depth,in sparseshippedbasins,or at highseastatesshowindications of twodistinctregimes associated withtheoccurrence ofbreakingwavesand a variationof root-mean-square pressure withthesquareof thelocalwind speed.In a previous paper[W. Carey,L Acoust.Soc.Am. Suppl.178,SI ( 1985) ], thegeneration of soundby waveturbulenceinteractionat low seastates(WS < 10m/s) andcollectivebubbleoscillations drivenby turbulenceat highseastateswasproposed (WS > 10m/s). Thispapershows thatrecentlyobserved bubblecloudsthat penetrateto tensof metersbelow the surfaceof the searesultin regionsof low sonicvelocitydescribed by Wood'streatmentof air-bubblewater mixtures,but with a density closeto that of water.It is shownthat sucha regioncan be treatedas a flexiblebody with mixture speedand density.The radiationfrom sucha bodywouldbemonopole anddipolein nature;butdueto theproximityof the seasurface,only the resultingdipolesourcewouldbe of importance. Theseregionscouldalsohavea resonant characteristic, andwhendriven by the turbulenceresult in sufficientradiated sound.These resultsare similar to and consistentwith the analysisby A. Prosperetti [ Nato AS 1, "Natural Mechanisms of Surface Generated Noise in the Ocean" (1987)1. at thesoundchannelaxisandwereuniformlyspacedat half-wavelength eitherat 200 or 300 Hz. From 75 to 300 Hz, the changein the vertical distributionof noiseshifts,especiallyat higherfrequencies, with increasingrangefromcoastalshippingwasin a mannerconsistent withtheeffect ofchemicalabsorption onlow-anglenoiseduetocoastalshipping. Whereasa shippingnoisepedestalat low anglesisobserved at all frequencies at shortrange,at longrangetheabsorption effectmakesthe verticaldistributionof ambientnoisemoreisotropicat higherfrequencies thanthat at lowfrequencies. Effectsof windspeedontheverticaldistribution of noise werealsomeasured.As wind speedincreases, ambientnoiseat angles greaterthan 15øfromthe horizontal(that is, noiseof localorigin) increases andapproaches thelevelsof thepedestalat 124øW andpresumablyexceeds themat highwindspeeds. Our resultsareorthogonal to those obtainedbyKibblewhiteetal., ARL/UT, withtheirdeployment at 150W of threeVedabsbouyson a north-southline to measurelow-frequency absorption usingnorthernshippingasa soundsource.Thisconflictsuggests a simultaneous north-south andeast-west experiment aswellasthe utility of usinga multielement verticalDIFAR array. [Work supported by ONR. ] 1:35 DD3. Estimation of surface noise source level from low-frequency seismoacoustic ambient measurements. Henrik Schmidt Engineering 5-204, Cambridge, MA 02139) and W. A. Kuperman (Naval ResearchLaboratory,Washington,DC 20375) The extraction 1:20 noise (MassachusettsInstitute of Technology, Department of Ocean of noise source levels from ambient noise measure- latitude of 32N at three stations, 124W, 136W, and 150W, the latter sta- mentsrequiresaccountingfor the seismoacoustic propagationfrom the surfacegeneratednoisesourcesto the field measurement positions.At frequenciesbelow 100 Hz, the waveguidenature of the environment stronglyinfluencesthe distributionof ambient noise.Only when these propagation effectsarc considered canmeasurements from differentexperimentstakenin differentenvironments be combinedto ascertainthe globalpropertiesof surfacenoisesources. Here, sucha previousanalysis of a shallowwaterexperiment[Schmidtet al., in NaturalMechanisms of SurfaceGeneratedNoise,editedby B. Kerman (Reidel, Dordrecht, The tionbeingabout1700milesfromthecoast.The arraysusedweredeployed Netherlands, 1988)] is usedand the resultsare combinedwith Kibble- DD2. Effect of longitudeon vertical distributionof ambient noisedue to wind and coastalshipl•ing.F. H. Fisher and W. S. Hodgkiss (Marine PhysicalLaboratory,ScrippsInstitutionof Oceanography,Universityof California--SanDiego,La Jolla,CA 92093) Using multielementvertical arrays deployedfrom FLIP, measurements of the vertical distribution $62 of ambient noise have been made at a d. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America $62 whiteandEwans'lower-frequency results[J. Acoust.Soc.Am. 78, 981994( 1985) ], thelatteralsocontaining a summary ofprevious experimental resultsobtainedby others.Thesedataaretreatedin thesamewayby accounting for theenvironments usinga wavetheorymodelof distributed noise[W. A. KupermanandF. Ingenito,J. Acoust.Soc.Am. 67, 19881996(1980)]. Whentheparticularenvironments areused,thespreadin thereprocessed noiselevelsissubstantially reduced,showingmoreconsistencyin the frequencydependence of the noisesources, suggesting that thereareonlya few (or possibly onlyone) dominantnaturalnoisesource mechanisms that primarily contributeto oceanambientnoisebelow 100 Hz. The importanceof appropriately includingpropagation factorsin processing noisedatafor estimating sourcelevelsis thereforestrongly demonstrated. 1:50 DD4. Downslopepropagationand verticaldirectionalityof windnoise. W. M. Carey (Naval UnderwaterSystemsCenter, New London,CT 06320), R. E. Evans(ODSI, Stonington,CT 06378), and J. A. Davis (PSI, New London, CT 06320) wind speedand rain rate, the soundgenerationmechanisms remainobscure.This isduepartlyto thedifficultyof simultaneously measuringthe sound-generating phenomenaat the locationof the acousticmeasurements.Also,thereisa lackof frequency spectraof soundgenerated bythe specific sources thatmaydistinguish betweenwind-andrain (or spray)generatedsoundandmay holdimportantcluesto the generatingmechanisms.A pier facility was established on NarragansettBay to monitor broadbandsoundassociatedwith observedseasurfacephenomena.Simultaneous recordingof windvelocity,waveheight,andrain rateallows precisecomparisonof the environmentalvariableswith their acoustic response. In addition,visualobservation is madeof whitecapping and of interfering shiptraffic.Theambientsoundspectraareroutinelygenerated ashigh-resolution FFT estimateson a MASSCOMP computer.Preliminarymeasurements showsoundpressureto vary linearlywith wind speed, withcorrelations rangingfrom0.88-0.97.The slopechanges aroundwind speeds of 6-7 m/s, i.e.,wherewhitecapsform,The spectraof wind-generatedsounddisplaya fasterroll-offthan rain-produced sound,the latter exhibitinga broad peakaround 15-18 kHz. Plotsare presentedof the intenseinsonificationof the water column by rapidly advectingrain squalls.[Work supportedby NAVSEA/AIRSEA systemscommands.] Measurements of theverticalnoiseintensityversus angle[ W. Carey and R. Wagstall',J. Acoust. Soc. Am. 80, 1523-1526 (1986)] show the low-frequency distribution(4200 Hz) to be broadlypeakedaboutthe horizontal,whereasthe higher-frequency (m400 Hz) distributionis peakedat theSOFAR angles( m + 15) neartheaxiswith a minimaat the horizontal. Thiseffecthasbeenattributedtothenoisefromshipsoverthe basinmargins[R. Wagstaft, J. Acoust.Soc.Am. 69, 1009-1014( 1981) ]. However,sincespectralvariationalongthe horizontalis smoothand the effectis observed in sparselyshippedareasof the world,windnoisemust bc includedto explainthiseffect.R. W. Bannister[J. Acoust.Soc.Am. 79, 41-48 (1986) ] hasattributedthiseffectto high-latitudewindsandthe shallowingsoundchannelfound in SouthernHemispherewaters.W. Carey [J. Acoust.Soc.Am. 79, 49-59 (1986)] attributedthe effectto downslopepropagation.Calculationsbetween50 and 400 Hz of the midbasin vertical directionalitymade with ASTRAL and PAREQ with a geoacoustic modelwereboth foundto showthat the bottombehavesasa low-passfilterasthelow-frequency energyat higheranglesinteractswith thebottomandisconverted to low-angleenergyin thedeepsoundchannel while,at the higherfrequencies andhigherangle,energyis absorbed. This low-passeffectis consistent with experimentalresults.Propagation of low-grazingangleenergyfromtheshelfis prominentin thecomputationalresultsbut not apparentin measured data,perhapsdueto bathymetricroughness. 2:05 2:35 DD7. Observationsof deep-oceanambient noise using a seismometer array. AnthonyE. Schreiner,LeRoy M. Dorman,JohnA. Hildebrand, Dalia Lahav (Scripps Institution of Oceanography,University of California-SanDiego,La Jolla,CA 92037), and Dale Bibee(NORDA, NSTL Station, MS 39529) An arrayof 12ocean-bottom seismometers (OBS) wasdeployedthis pastMarchin thedeepocean nearSanDiego.Thearrayhada maximum extentof 150m anda minimuminterelement spacingof 8 m. The accuracy of the relativesensorpositionswas as goodas 2 m in somecases.The seismometers weredeployedin 3.8km of waterwitha partiallymaneuverablevehicleat the end of a wire guidedby a transponder net. The OBS capsules containeda verticalandtwo horizontalseismometers anda hydrophone. Threeof theinstruments hada low-frequency hydrophone of theCoxdesign.Three-minwindowswererecordedat intervalsof 3 or 6 h forupto 30days.Spectrallevelshavebeencalculated between 0.01and30 Hz. A powerlevel drop of up to 40 dB belowthe microselampeak at frequencies below0.1 Hz wasobserved. Similarspectrahavebeenreported by Webband Cox [J. Oeophys.Res.91, 7343-7358 (1986) ]. At the time of the abstractdeadline,correlationwasbeingcalculatedas a function of sensorseparation to determinethecorrelationlengthof the ambient _noise. DDS. Underwaternoiseproducedby rainfall. H. C. Pumphreyand L. A. Crum (National Center for PhysicalAcoustics,The University of Mississippi,Oxlbrd, MS 38677) When a liquid drop falling throughthe air strikesa fiat horizontal surface,thereareacoustic emissions associated withthisimpactin both theair andtheliquid [ G. J. Franz,J. Acoust.Soc.Am. 31, 1080(1959) ]. Undercertainconditions, a smallair bubblecanbeentrainedin the liquid by theimpactingdropandstimulatedinto volumepulsations that radiate quite strongly.When severaldrops impact the surfacein a short time interval as in rainfall, the acoustic emissionsassociatedwith the bubble oscillations appearto bea majorcontributionto the total acousticemissionovera considerable bandwidth.With artificiallycreatedrainfall, the spectralpeakobserved near20 kHz for naturalrainfallcanbeduplicated [Scrimgeret al.:,J. Acoust.Soc.Am. 81, 79 (1987) ]. However,with the additionof surfacetensionreductionagents,thispeakissignificantly reduced,implying that a major portion of the noiseassociatedwith rainfall is related to oscillations of the entrained bubble rather than due to the impact of the drop with the liquid surface.[Work supportedby the ONR. ] 2:20 DD6. Correlation and spectra of ambient sound generatedby wind, whitecaps, and rain. David Shonting (Naval Underwater Systems, Center, Newport, RI 02841), Foster Middleton (University of Rhode Island, Kingston, R1 02881), and Nancy Taylor (Applied Science Associates, Narragansett,RI 02882) Althoughhighcorrelationexistsbetweenoceanicambientsoundand $63 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 2:50 DDS. Ambient ULF/VLF ocean-bottom noise 200 km west of San Francisco: Motion and pressure 0.002 to 20 Hz. G. H. Sutton, J. A. Carter, and N. Barstow (Rondout Associates,Incorporated,P.O. Box 224, StoneRidge, NY 12484) Data recordedfrom long-periodand short-period,three-component seismometers and hydrophones of the Columbia-PointArena ocean-bottom seismicstation(OBSS) are beinganalyzedin the frequencyrange 0.001 to 40 Hz to determine the characteristics of ULF/VLF ocean-bot- tom noise. The OBSS, located at 38ø 09.2' N, 124ø 54.4' W at 3903-m depth, operatedfor over 6 years,from May 1966 to September1972. Selected portionsof essentially continuous FM tapedataarebeingdigitized to obtainULF/VLF spectraand covariances duringquietand noisy times and during passageof vesselsand earthquakewavetrains.It is plannedto compareOBSSdatawith availablesourceinformationon sea/ swell conditions, tidal currents (measured by a current meter on the OBSS), earthquakes,and shipping.Spectrallevelsappear to be above systemnoisefor mostfrequencies aboveabout0.002 Hz. The pressure spectraagreewith thoseobtainedmore recently [Cox et al.,Atmos. OceanicTech. 1, 237-246 (1984) ], showinga strongminimumbetween about0.03 and 0.1 Hz. The motion spectrafrom the OBSS seismometers are similar.Coherencybetweenverticalmotionandpressureiswell above random for most of the spectrumbelow 0.5 Hz. There is clear evidence that ambient noise maxima rise and fall with weather conditions and wa- ter waveactivityand that ULF noise,especiallyon the horizontalseismometers,hasa tidal dependence.[Researchsupportedby ONR.] 114thMeeting:AcousticalSocietyof America S63 WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 TUTTLE SOUTH ROOM, 1:30TO 3:00 P.M. SessionEE. Noise IV and EngineeringAcousticsIII: Applicationof Math Models and Experimental Investigationsto SolveAcousticProblems(PosterSession) Larry H. Royster,Chairman 1217BR Hall, M•4E Department,North CarolinaState University, Raleigh,North Carolina27695-7910 ContributedPapers All posters willbedisplayed from1:30to 3:00p.m.To allowcontributors theopportunity toseeotherposters, contributors of odd-numbered paperswill beat theirposters from 1:30to 2:15p.m.andcontributors of evennumberedpaperswill beat theirpostersfrom2:15to 3:00p.m. EEL Boundary elementsolutionfor a coupledelastodynamicand wave equation system to predictforcedresponse of a plugged acoustic cavity. Robert Ciskowski (International BusinessMachines, Rochester, MN 55901) andLarry H. Royster(Departmentof MechanicalEngineering, North CarolinaStateUniversity,Raleigh,NC 27695-7910) The subjectof this investigation is the useof the boundaryelement method(BEM) to predictthe steady-state and transientresponse of a plugged acoustic cavity to external excitation. The plug is an clastodynamic materialsubjectto externalexcitation.The excitationof theplugistransmitted to theacoustic cavity.No internalsources of excitationare considered. A three-dimensional BEM modelis developedfor theanalysis,bothfor thefluidin thecavityandtheclastodynamic material in contactwith it. Transientresponse is determinedby recoveringthe timedomainresponse from the Laplacetransformdomainsolutionusing conditionrequirementson the acousticvariables'valuescomposingan acousticpulsewill beaddressed and resultsfor the numericallysimulated propagationof sucha pulsewill be shown.Further, an extensionto the abovemodelfor atmosphericpropagationcurrentlybeingtestedwill be described that includesbothclassical and molecularrelaxationabsorption effects.Spacepermitting,the implementation of our programson a CRAY X-MP supercomputer will becommented upon. EF_A.An on going study of the use of the boundary element method to solve some of the more commonly encountered industrial noise and an FF"r-basedinversionscheme.Steady-stateresponseis obtainedfrom the Laplacetransformdomainsolutionby settings = -- iw. The numeri- vibration control problems. Kassem M. Mourad (Department of MechanicalEngineering,North CarolinaStateUniversity,Raleigh,NC calcodeis designed to run on theIBM familyof personal computers and 27695-7910), Robert D. Ciskowski (International BusinessMachines, workstations. Rochester,MN 55901), and Larry H. Royster (Department of MechanicalEngineering,North Carolina StateUniversity,Raleigh, NC 27695-7910) EE2. Transmission loss studies of muffler and duct systemsby the boundary element method. C. Y. R. Cheng and A. F. Seybert (Department of Mechanical Engineering, University of Kentucky, Lexington,KY 40506-0046) The presentpaperdealswith transmission lossstudiesof acoustical mufflerand duct systemsusingthe boundaryelementmethod(BEM). Resultsobtainedby usingthis approachare presentedfor both axisymmetric and fully three-dimensionalmodels.The transmissionlossof a mufflerelementcan be computedusingthe transfermatrix approach. A studyis underway to establishthe advantages or disadvantages of usingthe boundaryelementmethod(BEM) to obtainsolutionsto some of themorecommonlyencountered noiseandvibrationcontrolproblems encountered in generalindustry.Initial investigations includepredicting the steady-stateand transientroom-volumeSPL responsefrom internal and boundaryacousticsourcesfor variousboundaryimpedances. The numericalcodedeveloped to obtainthesolutions of suchproblems was designed to run on the IBM familyof personalcomputers and workstations. This method is ideally suited for the BEM becauseonly the valuesof the unknownsat the inletandoutletboundaries are required.The BEM valuesare comparedto theoreticalpredictionsand other availabledata such as finite clementdata, and goodagreementis obtainedwithin a broad frequency range. It canbeconcluded thattheBEMshould givereliable estimates of transmission lossin practicalsituations. [Work supported in part by ASHRAE. ] EE5. Plane-waveexcitationof an infinite circularcylinderreinforcedby a periodicset of rings.FrancoisDe Maigrct (Thomson-SintraA.S.M., Route desDolines, Parc de Valbonne,B.P. 38 06561, ValbonneCedcx, France) Analyticalexpressions are derivedfor the scatteredfieldby an infinite EE3. Finite differencemodelingof acousticpulse propagationin the atmosphere.Victor W. Sparrow (Department of Electrical and Computer Engineering,University of Illinois, Urbana-Champaign, Champaign, IL 61820andUSA-CERL, P.O.Box4005,Champaign, IL 61820-1305)andRichard Raspet(PhysicalAcousticResearchGroup, The University of Mississippi,University, MS 38677) In a previouspresentation [J. Acoust.Soc.Am. Suppl.1 g0, S104 (1986)], theuseof a finitedifference approximation of thetime-dependentacoustic equations forarbitraryaxisymmetric pulses wasdescribed. In thatpaper,however,onlytheapplication of theseequations for a time harmonicsimplesourcewasdemonstrated. In thepresentation, theinitial S64 J.Acoust. Soc.Am.Suppl. 1,VoL82,Fall1987 fluid-loaded cylindricalshellreinforced withtheperiodicsetof ringsand acousticallyexcitedby an incidentharmonicplanewave.Vibrationsof the shellare treatedusingthin bendingcylindricalshelltheoryby Kennard. The stiffenerainteract with the cylindricalshell only throughnormal forces.Unknownquantities(i.e., scatteredacousticfield,displacements of theshell)areexpandedon the naturalBloch-Floquetbasisinducedby thestructureperiodicity.Numericalsimulations ona widerangefrequen- cy are reported,wherethe dimensionless product"ka" of the incident fluidwavenumber bytheshellradiustypicallyextends from10-2 to 102, usinga uniquealgorithm.Numericalpredictionsare comparedwith experimentaldata on differentphysicalsituationsshowinga goodagreement. [Work supportedby GERDSM.] 114thMeeting: Acoustical Society ofAmerica S64 EE6.Effectof background noi•eonsound powerestimates usingthe EE9. Effect of a •emicircular diffuser on the soundfield in a rectangular soundintensitytechnique.U.S. Shirahatti,Malcolm J. Crocker,and P. K. Raju (Departmentof MechanicalEngineering,Auburn University, Auburn, AL 36849) room.Pan Jie and D. A. Bies(Universityof Adelaide,Departmentof MechanicalEngineering,G.P.O. Box No. 498, Adelaide,S.A. 5001, Thepresence ofa background noisesource affects theaccuracy of the estimateof the •ourcesoundpowerlevel.In thispaper,the resultsof a systematic experimental studyon the accuracyof estimation of sound powerlevelsin thepresence ofa background noisesourcearcreported.A calibratedsoundpowersourcewasusedin theseexperiments. The sound powermeasurements weremadeusingthetwo-microphone soundintensitytechnique. Theerrorsin theestimates of thesoundpowerlevelof the Previousexperimental workhasindicatedthe influence of a rotating diffuseruponthe diffusionof a reverberation field,uponthe radiation impedanceof a soundsourcein a reverberation room, and uponthe boundaryabsorptions. Someinteresting resultsofthisworkneeda quantitativeinterpretationwhileotherssuggest an effectiveinvestigation to obtain a detailedphysicalinsight.In an effortto findan insightinto andan interpretation of thiswork,an analyticalapproach is presented. Thispaperisan interimreportoncontinuingresearch, andit reviewsan investigationof the effectof a semicirculardiffuseruponthe two-dimensional soundfieldin a rectangularroom.The numericalrelationship betweenthe resonance frequencies oftheacoustical modesin theroomandthediffuser orientationis given.The sound-prcssurc distributionsof the acoustical modesarc alsogivenasthe functionof the diffuserorientation.A thrcedimensionalcavityis usedfor experimentalverification.In this cavity, onlythesoundwavesin thehorizontalplanewill beaffected bya diffuser. The experimentaland numericalresultsagreeclosely. calibrated sound source were determined for different levels of back- groundnoise.Also,a systematic studywasmadeon the relationship betweenthe numberof pointsof measurement on the enclosing surface definedaroundIhe sourceandtheaccuracyof thesoundpowerestimate. It hasbeendemonstrated that the accuracyof soundpowerestimatesis greatlyimprovedby continuous handscanning of theenclosing surface insteadof by makingmeasurements at a smallor moderatenumberof fixedpoints.Thelocalsoundpressure minusintensityindexandtheglobalsoundprcssu rcminusintensityindexarcusedasindicators of thedata quality.[Work supported by IBM, Charlotte,NC.] EE7. Acousticradiation from plates driven by multiple point random forces.H. Peng and R.F. Keltie (Center for Sound and Vibration, Departmentof MechanicalandAerospaceEngineering,North Carolina StateUniversity,Raleigh,NC 27695-7910) The problemof acousticradiationfromplatesdrivenby multipleran- dompointforcesis analyzedin thisstudy.First,theformulation of the soundpowerradiatedby infinitethin plates(one dimension)underthe actionof multiplerandompointforcesis derived.The reactionof the acousticmediumon thevibratoryresponse of platesistakenintoaccount. With theinputspectrumbeingassumed asbandlimitedwhitenoise,the effectsof input forcecorrelationon the radiatedsoundpowerlevelare examined.The analysisis then extendedto finite thin plateswith four simplysupported edges. Thefluidloadingeffects areneglected. The exact expressions for thesurfaceacoustic intensityandtheradiatedsoundpower arealsoderived.Assuminglight damping,approximate solutions for acousticintensityand soundpowerareobtained.An experimental study is carriedout to measurethe surfaceacousticintensitypatternsand com- parewith theanalyticalresults.The effectsof inputforcecorrelationon theacousticintensitypatternsandthesoundpowerlevelarediscussed. EE8. Prey detectionby meansof passivelisteningin bottlenosedolphins {Tursiopstruncatus}. N61io B. Burros and Arthur A. Myrberg' Jr. (RSMAS-BLR, Universityof Miami, 4600 RickenbackerCauseway, Miami, FL 33149-1098) Australia) EEl0. The applicationof thecepstrumtechniqueto theseparationof the input signaland the structureresponse.Mei Q. Wu and Malcolm I. Crocker (Department of Meehanical Engineering,Auburn University,Auburn, AL 36849) In thispaper,aniterativealgorithmfortheeepstrum ispresented. The improved cepstrum technique isappliedtoseparate theinputsignalfrom theresponse of a structure. It isprovedthatif theinputsignalisan impulseandthefrequency response ofthestructure isbandlimited, thenthe frequency response function of thestructure canbedetermined fromthe outputof thestructure withoutanyprcknowledge of the inputsignal. Furthermore, thespectrum oftheinputsignalcanbedetermined fromthe structureresponse functionand the output spectrum.Somecomputer simulated experiments havebeenconducted. The experimental results showthatthefrequency response functionrecovered byusingtheiterative algorithmconverges to thecorrectresponse function,andfor about30 timesof iterationthe recovered response functionis verysimilarto the correctresponse function.[Worksupported byIBM, Austin,TX. ] EEII. Subjectiveevaluationof a test earmuffexhibitingfiat attenuation withnonlinearcharaeterlsties. JuneL. Peters(Departmentoflndustrial Engineering, North CarolinaStateUniversity,Raleigh,NC 27695), Larry H. Royster(Departmentof MechanicalEngineering, North CarolinaStateUniversity,Raleigh,NC 27695), JuliaDoswellRoyster ( EnvironmentalNoiseConsultants,Inc., Cary, NC 25712), and Richard Free-livingbottlenosedolphins(Tursiopstruncatus)are primarily piscivorous andtheuseof activeecholocation hasbeenimplicatedin prey capture[K. S. Norris and B. M6hl, Am. Nat. 122, 85-113 (1983)]. Re- G. Pearson(Departmentof IndustrialEngineering, North Carolina cent evidence ba.•xt on the examination of stomach contents of more than Subjective userresponses wereobtained for a testearmuffexhibiting approximately flat attenuation of about25 dB from 500-8000Hz for 70 individualsfrom the southeastern United Stateshassuggested that bottlenosedolphinsalsolikely use passivelisteningfor the purposeof detectingandpossibly orientingto theirprcy.The mostfrequentlyoccurringpreyspecies (i.e.,thosefoundin morethan ! 5% of thesample)werc invariablythoseknownto be highlyactivesoundproducersand, in all instances, producersof soundsof highlevel(e.g., croakersanddrumsof the family Sciaenidae, gruntsof the family Haemulidae,toadfishes-midshipmenof thefamilyBatrachoididae, andmulletsfromthefamilyMugilidae).Althoughin severalinstances members of somespecies couldhave beencapturedby activeecholocation and visualmeans,it ishighlyunlikely that membersof otherspecies, by their highlysecretivehabits,would havebeencapturedby suchmethods(i.e., membersof the family Batrachoididae).Data from other regionscorroboratethe abovementioned findingsthat fishproducingloud soundcomprisethe bulk of the diet of bottlenosedolphins. S65 J. Acoust.Soc. Am. Suppl. 1, VoL 82, Fall 1987 StateUniversity,Raleigh,NC 27695) SPLs lessthan 115 dB, with attenuation increasingnonlinearly up to about10 dB at higherSPLs.The studypopulationconsisted of police officers performing annualgunfirerequalification, in whichofficers executerequiredtargetshotsduringtwo sequential relayexercises. In the firstphaseof thestudy,subjects woreeithertheirregularhearingprotector or thetestmuffduringthefirstrclay,thenthealternateprotectorfor thesecond rclay( withordercounterbalanced). In thesecond phase, additionalsubjects woreeithera controlearmuffthatwasidenticalto thetest earmuffexceptfor the nonlinearmechanism(alsoeliminatingthe flat response characteristics) or the testearmuff.Subjects completed questionnaires concerning theiropinionsof thehearingprotectors wornduring both phases.Preliminaryresultsindicatea significant differencein favor of the test earmuffin threeareas:speechunderstanding,perceived levelof protection, andcomfort. 114th Meeting:AcousticalSocietyof Amedca S65 farfieldsolution.The nonlineareaseof strongscatteringin the inhomogeneous mediumis considered. The Feynmanpathintegralis appliedto theproblem,andthiscangivethesolutionin closedform.Previousworks [M. SlaneyandA. C. Kak, paperat the 1985IEEE Ultrason.Syrup.;and Z. Q. Lu, IEEE Trans.Ultrason.Ferroelectries andFrequencyControl UFFC-33(6), 722-730 (1986) ] cannotgivethe solutionin closedform. The pathintegralisbasedon theactionprinciple(variationalprinciple). Here, the Schr'ddinger equationwill be used.The scatteringintegrals, EEl2. A generalized theoryfor acoustical holography and acoustical diffractiontomography. W. S. (}an (AcousticalServicesPte Ltd., 29 TelokAyerStreet,Singapore 0104,Republicof Singapore) A generalzied theoryisproposed tocombine acoustical diffraction tomography(ADT) withacoustical holography (AH) totakeaccount ofthe diffraction andscattering duringtransmission ofsoundwavesthroughan inhomogeneous medium.Thisis usuallyneglected in AH but mustbe considered if themediumis inhomogeneous. In thisgeneralized theory, perturbation expansion of thescattering potential,andthewavefunction are all expressed in termsof the path integrals.The algorithmof the imagingprocess for the generalizedtheoryis given. AH is a nearfieldsolutionof the HelmholtzwaveequationandADT is a WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 UNIVERSITY LECTURE HALL, 2:15 TO 3:15 P.M. Session FF. Architectural Acoustics IV: Technical Committee on Architectural Acoustics V. O. Knudsen Distinguished Lecture William J. Cavanaugh,Chairman Cavanaugh TocciAssociates, Inc., 327 F BostonPostRoad,Sudbury,Massachusetts 01776 Chairman's Introduetion--2:15 Invited Paper 2:20 FFI. Designingfor the performingarts:An historicaloverview.Michael Forsyth(Departmentof Drama, Universityof Bristol,29 Park Row, BristolBS1 5LT, United Kingdom} A relationship developed historically between theacoustics of differentauditorium typesandcorrespondingstyles ofmusical composition. Thisresulted fromthecornposer's intuitiveawareness ofroomacoustics, and fromthe architect's limitedknowledge, basedon trial and error,of the acousticperformance of building materialsandroomshapes in providingfor differentpurposes. Exceptwherecomposers havehadauditoria specifically designed for theirownmusic,thisrelationship becametenuous in the presentcenturywhenthe musicalrepertoire broadened andwhennonmusical criteriacameto dominate thedesignprocess. An auditorium'ssuccess nowdepends ontheinitialselection ofacoustical andothercriteriain relationtothebuilding's defined purpose andonthetranslation ofacoustic theoryintothree-dimensional form.Thesefactors together determine theaudience-performer relationship andtheshapeanddimensions oftheenclosing surfaces. Historical examples illustrateearlyattemptsto projecttheoryinto built form,and othersfrom the presentday describe theproblemof integratingopposing acousticdemands. S66 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 114thMeeting: Acoustical Society ofAmerica S66 WEDNESDAY AFTERNOON, 18NOVEMBER 1987 ORCHID B ROOM, 2:00P.M. Meeting of Accredited StandardsCommittee S2 on Mechanical Shock and Vibration to be heldjointly with the TechnicalAdvisoryGroup(TAG) Meeting for ISO/TC 108 MechanicalVibration and Shock J. C. Barton, Chairman S2 CaterpillarTractorCompany, Research Department,100N.E. Adams,Peoria,Illinois61629 G. Booth,Chairman,TechnicalAdvisoryGroupfor ISO/TC 108 220 Clark Avenue,Brandford,Connecticut 06405 Standards CommitteeS2 onMechanicalShockandVibration.Workinggroupchairswill presentreportsof theirrecentprogress onwritingandprocessing various shockandvibrationstandards. Therewillbea reporton the interfaceof S2 activitieswith thoseof ISO/TC 108 (the TechnicalAdvisoryGroup for ISO/TC 108 consists ofmembers ofS2,S3,andotherpersons notnecessarily members ofthosecommittees) including plans for thenextmeetingof ISO/TC 108,tobeheldin September 1988. WEDNESDAY AFTERNOON, 18 NOVEMBER 1987 UM AUDITORIUM, 3:30 TO 5:00 P.M. Plenary Session Chester M. McKinney, Chairman President,AcousticalSocietyof America Presentation of Awards DistinguishedServiceCitation to FrederickE. White HonoraryFellowshipto Henrik A. S. N4dtvedt SilverMedal in Psychological and Physiological Acousticsto EberhardZwicker SilverMedal in SpeechCommunicationto Dennis H. Klatt Trent-Crede Medal to Miguel C. Junger Local musical entertainment S67 J.A,coust. Soc.Am.Suppl. 1,Vol.82,Fall1987 will conclude the session. 114thMeeting: Acoustical Society ofAmerica S67 THURSDAY MORNING, 19 NOVEMBER 1987 POINCIANA ROOM, 8:30 TO I 1:50 A.M. SessionGG. Musical AcousticsI: Extended Vocal Techniques,and Production and Perceptionof Music Gary L. Gibian, Chairman Department of Physics and•4udioTechnology, •4merican University, Washington, DC 20016 Chairman's Introduction---8:30 Invited Paper 8:35 GG1. Harmonic singing and the Harmonic Choir. David Hykes (The Harmonic Arts Society, 1047 Amsterdam Avenue, New York, NY 10027) In harmonicsinging,thesingeremphasizes a selected upperharmonicofthevocalpulse.In thisway,hecan singtwo notesat once,thefundamental andtheselected harmonic.Membersof the HarmonicChoirhave developedtechniques for the followingeffects:( l ) The fundamentalis constantwhile the selectedharmonic varies;(2) the fundamentalvariesin a melodywhile the selectedharmonicremainsconstant,leadingto parallelharmony;(3) boththefundamental andtheselected harmonic vary,eitherin converging or in divergingdirections. TheChoirtogether performs unaccompanied workswhicharecomposed butnotscored. Differentperformances followa commonpathbut differin manydetails. ContributedPapers 9:20 9:50 GG2. Regional cerebral blood flow for singers and nonsingerswhile speaking, singing, and humming a rote passage. C. Formby GCr4. Modal analysis of a Caribbean steel drum. Uwe I. Hansen (Departmentsof CommunicativeDisordersand Neurology,University of Florida, Gainesville,FL 32610) and R. G. Thomas(Departmentof Biometry,EmoryUniversity,Atlanta, GA 30322) Two groups of singers(n = 12,13) and a group of nonsingers (n = 12) eachproducedthe nationalanthemby (1) speakingand (2) singingthe words,and by (3) hummingthe melody.Regionalcerebral bloodflow (rCBF) wasmeasured at restandduringeachphonationtask from sevenareasin eachhemisphereby the 133-Xe-inhalationmethod. Global,intrahemisphere, andinterhemisphere rCBF weregenerallysimilar acrossphonationtasks and did not yield appreciabledifferences amongthe nonsingers and the singers.From theserCBF data, it was concluded that: ( 1) thenormalproductionof a familiarpassage by speaking, singing,or hummingrequiresthe interactionof bothcerebralhemispheres moreor lessequallyand (2) thesetasksarerelativelyindependent (Department of Physics,Indiana State University,Terre Haute, IN 47809) and Thomas D. Rossing (Department of Physics,Northern Illinois University,DeKalb, IL 60115) Modesof vibrationand couplingbetweenadjacentnote areasin a double-second steeldrum are comparedby severalcomplementary techniques,includingimpactmodalanalysis,holographicinterferometry,and recordingsoundspectraundervaryingconditionsof damping.Eachnote has at least one overtone mode tuned to a harmonic of the fundamental frequency. StrikingtheA• notearea,forexample, excites notonlya secondharmonicA• in thatsamenoteareabut alsothestrategically located A} andA} noteareasadjacent to it. Thecoupling depends uponamplitudein a nonlinearway. of musicaltraining. [Researchsupportedby NIH. ] 10:05 9:35 GG3. Note coupling in Caribbean steel drums. D. Scott Hampton, Clifford Alexis, and Thomas D. Rossing (Department of Physics, Northern Illinois University,DeKalb, IL 60115) The acousticalbehaviorof Caribbeansteeldrumsis characterizedby strongcouplingbetweenthe differentnoteareasIT. D. Rossirig,D. S. Hampton, and J. Boverman,J. Acoust. Soc. Am. Suppl. I 80, S102 (1986) ]. This makesit difficultto maintaintuningstability,especially in the harmoniesof the variousnotes.An improvednewdesignmakesuseof doublegroovesto separatethevariousnoteareas.By meansof holographic interferometry,the note couplingwascomparedin two comparable doublesecond(alto) drums,onewith singlegroovesandonewith double grooves. The drumwithdoublegroovesexhibitslessinternotecouplingat highamplitudes,andbettertuningstability. S68 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 GGS. Physical correlates of perceptual similarity among synthesized approximations to selected targets. G. L. Gibian, D. R. Clements, E. N. Hamden, H. E. F. Williams, and R. K. Massaro (Physics Department,AmericanUniversity,Washington,DC 20016) Experimentsare describedto evaluatethe relativemeritsof several measures for predictingthe perceivedcloseness of synthesized approximationsto selectedsteady-state portionsof musicaltones("targets"). Approximations weresynthesized bymeansof audio-rate frequency modulation[Crowning,J.AudioEng.Soc.21,526-534( 1973) ] usinga computer programdevelopedin a previouspaper [Gibian et al., J. Acoust. Soc.Am. Suppl. 1 81, S46 (1987), and Audio EngineeringSocietyPreprint #2380]. The importanceof includinglevel-and frequency-dependent error weightingsaccordingto the Fletcher-Munsoncurveswill be assessed. This informationcan be usefulto composerswho are blending electronic and traditional instruments in mixed ensembles. 114thMeeting:Acoustical Societyof America $68 10:20 GG6.Comprehensive studyofanalysis andsynthesis oftones byspectral bowpressure, follows asa special case.[Worksupported in partbyNSF and the French Ministry of Culture.] interpolation. RogerDannenberg, Marie-HcleneSerra, and Dean Rubin (Center for Art and Technology,CarnegieMellon University,Pittsburgh,PA 15213) A newapproach to thereal-time generation of digitalsounds usesa completely automated analysis/synthesis technique for naturalsounds. Thisapproach leadsto a moreefficient implementation thanclassical additivesynthesis; moreover it allowsdynamicspectralvariations to be controlled withonlya fewhigh-level parameters. Additivesynthesis devicesrequirea largenumberof oscillators (onefor eachpartial).This technique givesexcellent results; however,it requires a largeamountof computation, and a largeamountof controldata. On the otherhand, 11:05 GG9. Impedancecharacteristics abovecutofffor clarinet-likestructures. William J. Strong(Departmentof Physicsand Astronomy,Brigham YoungUniversity,Provo,UT 84602) Thecutofffrequency determined by theopentoneholesona clarinet- fixed-waveform synthesis usesonly one oscillator,but the resultsare of like structureis a usefulmeansfor characterizing thestructure'smusical poormusical qnalitysincethereisnodynamic evolution ofthespectrum. properties. [See,for example,thediscussion anddatain A. H. Benade, FundamentalsofMusicalAcoustics (OxfordU.P., London,1976),Chaps. 21-22; noteSec.22.5 in particular.]Whenimpedance peaksabovethe A new techniquehasbeeninvestigated in whichspectralvariation:is achieved through spectral interpolation. Theresearch shows thatspectral interpolation provides high-quality synthesis including controlled timbrai variationat littlemorethanthecostofa table-lookup oscillator. Thetask of analyzingdifferentkindsof instrumental soundsto producecontrol informationfor thistechnique hasbeenautomated. cutofffrequency haveamplitudes thataresignificant fractions of those belowcutoff,theymay influence theclarinetreedvibration.The input impedance of a clarinet-like structure(a cylindrical tubein thepresent case)is influencedby its inputandoutputterminations. Threeinput terminationsare considered: (1) rigid cap, (2) rigid taper,and (3) rigid taperwithreed.Several outputterminations areconsidered including ( 1) openend,(2) "infinite"toneholelattice,(3) "short"toneholelattice, and (4) shortlatticeplusbell.Input impedances arecalculatedfor variouscombinations of the inputandoutputterminations to demonstrate theireffects on theamplitudes of theinputimpedance peaksbothabove 10:35 and belowthe cutofffrequency. GG7.Nonuniformity in timbreof stringinstruments. Asbj•JrnKrokstad (Division of Telecommunications, The NorwegianInstitute of Technology, N.7034Trondheim,Norway) Mechanical musicalinstruments, mayshowgreatvariationof all physicalpropertiesthat correlateto timbre,evenoverintervalsof a semitone.To examinepreferences andlimitsofacceptability, timbrevariations havebeenstudied in a melodic context.A shortmelodyof 13tonescover- inga decime Wasplayed bytwoprofessional musicians onthreeviolins anda viola,andrecorded at two microphone positions in an anechoic room.Eachtonewasprocessed digitallyto reducedifferences in loudness andin transients. In thefirstoftwolistening tests,dissimilarities in timbre betweenpairsof successive toneswereevaluated.From thesedifferences, integrated differences fordifferent versions ofthewholemelodyarecalculated,anddifferences between versions. The possibility of identifying a certainplayer,atcertaininstrument ora certainmicrophone position may begivenasrelativenumbers. Thedissimilarity dataareusedforthesynthesisof newversions of the melodywith a rangeof nonuniformity in timbre.In thesecond listening test,acceptability of nonuniformity were evaluated. To studythedimensions andphysical variabilities oftimbreby sustained tones,multidimensional analysisof the subjective differences areperformed andseveral spectral variables fordescribing thedifferences are tested. 11:20 GGI0. Spectralcharacteristicsof the Helmholtz resonatordrivenby an air jet. Roset Khosropourand Peter Millet (Department of Physics, HamiltonCollege,Clinton,NY 13323) Motivatedby an interestin theactionof airjetson resonantsystemsin woodwinds, preliminarymeasurements havebeenmadeon thefrequency spectraof severalair jet driven Helmholtz resonators.Over the rangeof air jet velocitiesexamined,stableoscillations dominatedby a singlefrequency (under certain conditionsharmonicsare in evidence)occur in severalintervalsof jet velocity.Within eachsuchintervalthe dominant frequencyincreases with increasing jet velocity.The spanof frequencies associated witheachintervalincludes thecalculated Helmholtzfrequency and,dependingon the resonatorand the interval,canbeasmuchasan octave.The spectrumof thezonebetweentheintervalsis formedfroman overlapof the extrapolatedspectraof the adjacentintervals,i.e., it is markedby the appearanceof two frequencies, whereone is abovethe maximumfrequencyfrom the lower interval and the other is below the minimumfrequencyfrom the higherinterval. 10:50 11:35 GGS. Theoreticalrelationshipbetweenbow pressureand vibration amplitude,derivedfrom a sticking-slidingforce-velocitycharacteristic. Xavier Boutilion and GabrielWeinreich (Randall Laboratoryof Physics,Universityof Michigan,Ann Arbor, MI 48109) GGI1. Acousticalbehaviorof a bass drum. Thomas D. Rossirig The Helmholtzmotionfor bowedstringsmaybedescribed with the helpof a characteristic between thebow'sfrictionalforceandthestring two largemembranes may be set to equaltension,but moreoftenthe batterheadis tunedto a highertension.The soundspectrumcontains velocityat the bowingpoint. Althoughthis characteristichasneverbeen many peaksthat canberelatedto the vibrationmodesof a membrane.The dependably measured, someof its featuresareimpliedby players'commonexperience:. In general,the vibrationamplitudechangeswith bow pressure orbowvelocity.Therelationships between thesechanges andthe lowtensionof theheadsplusthestrongcouplingvia theenclosed air lead (Department of Physics, Northern Illinois University, DeKalb, IL 60115) The bass drum is one of the loudest instruments in the orchestra. Its to a largefrequency difference (about1«octave}betweenthetwo (01) local evolution of the characteristic will be establishedand discussed.The modesassociatedwith in-phaseand out-of-phasemotion of the heads. Sounddecayratesfor mostmodesareslightlygreaterwhenthetwo heads caseofperfectstickiness, in whichtheamplitude becomes independent of are at the same tension. S69 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 114thMeeting: Acoustical Society ofAmerica S69 THURSDAY MORNING, 19 NOVEMBER 1987 UNIVERSITY LECTURE HALL, 8:30A.M. TO 12:00NOON SessionHH. Psychologicaland PhysiologicalAcousticsVI: External Ear, Inner Ear, and Cochlear Implants William W. Clark, Chairman CentralInstitutefor theDeaf, 818 S. Euclid,SaintLouis,Missouri63110 ContributedPapers 8:30 HHI. Static and dynamic compliancesof the rat teetorial membrane under in vh•o-like conditions. Glenn H. Frommet (DAMPA A/S, 5690 Tommerup,Denmark) Three-dimensional solutions for propagating wavesin thefluid-filled gapbetween thetectorialmembrane(TM) andtheorganof Corti (OC) areobtainedandincludedin an existinganalyticalmodelof thecochlea. The three-dimensional solutions giveessentially thesamecouplingbetweenTM and OC as that previouslyfound with two-dimensionalsolu- Compliances wereexaminedqualitativelyin 17 rat cochleas prepared minutesafter death.Forceswereappliedperpendicularlyto the membranebya 10-/tin-thick microelectrode thatcouldalsovibrate.Sinusoidal displacements were10-pmpeakto peak.White andstrobelight illuminationwasused [G. H. Frommet,ActaOto-Laryng.94, 451-460 { 1982} ]. Staticdepression of the membraneat pointsbetweenthe spiralsulcusto the outerhair cellsproducedan elongatedhollow,whilealmostcircular depressions werenotedin the lip region.Dynamiccharacteristics were frequencydependent.Under 60 Hz, depression patternsweresimilarto the static ease.However, at 200 Hz, the regionbetweenthe sulcusand outer hair cellswasquite rigid and displacedat the iimbal attachment. Local vibrationsof the membranein the lip regionand over the sulcus were almost circular. The mechanical characteristics correlate with ul- trastructural studies.It is concludedthat static analysesalone cannot describe the viscoelastic behavior of the tectorial membrane. tions.However,including thethree-dimensional solutions givesriseto a newpropagating mode.With two-dimensional wavesin thegap,themodel exhibits threeducts(twosealee andtheinnersulcus)andtwopropagatingmodes. With three-dimensional solutions, thegapfunctions asa fourthduct.Consequently, a thirdpropagating modeappears. The flow patterns associated withthismodearedescribed, andthemode'ssignificanceis discussed [Work supportedby NSF.] 9:15 HH4. Quantificationof cerebrospinal fluid(CSF)contamination of scala tympani (ST) perilymph samples. Akira Hara, Alec N. Salt, Jogy Varghese, and RuedigerThalmaon (Department Otolaryngology, Washin. gtonUniversity, SaintLouis,MO 63110) of Previouslyit hasbeenreportedthatlevelsof mostaminoacidsin small perilymphsamples( < 100nL) takenfromST throughtheroundwindow (RW) ofguineapigsaredistinctlydifferentfromthosein CSF [A. N. Salt 8:45 andR. Thaimann,Adv.Otorhinolaryngol. (in press)]. Thecomposition HH2. Stiffnessof the tectorialmembrane in liveMongoliangerbils.J. of samples of largervolumemorecloselyresembles CSF. Now an analysis of aminoacidcontenthasbeenperformedas the samplevolumewith- J. Zwislocki, M. W. Serafini, andL. K. Cefaratti{Institute forSensory drawnthroughtheRW wassystematically variedfrom 100nL to 5 Research, Syracuse University,Syracuse, NY 13244-5290) G!ycine, serine, glutamine, and alaninc were measuredby HPLC {Physiological Pico-Tag,Waters}.Up to 200 nL, the glyeinecontentof ST perilymphwassimilarto that of scalavestibuli.With largervolumes, theglyeinecontentrapidlydeclinedtowardthatof CSF.Thesedatasuggestthat perilymphsamplestaken throughthe RW are contaminated with CSF if morethan200 nL is withdrawn.If ST perilymphis sampled The classicalmodel of hair-cell stimulation in the mammalian cochlea, accordingto which the tcetorialmembraneactsas a stiff anchorfor the cells'sterocilia,requiresthe stiffness of the membranein the radial cochleardirectionto be greaterthanthat of thesterocilia.This stiffness wasmeasured bymeans offlexible, calibrated micropipettes inserted deep intothemembrane at thelocation oftheouterhaircells.Theywereintroducedintothecochleathroughan openingin thelateralwallof thescala media,asdescribed previously [J.J.Zwislocki,S.C. Chamberlain, andN. B.Slepecki, J.Acoust.Soc.Am. Suppl.181,S6 (1987) ]. Thestiffness was determined bymoving themicropipettes radiallyandbymeasuring their througha smallfenestra inthethirdturn,largervolumes (upto2/•L) can bewithdrawnwithoutsignificant CSF contamination, evidentlybecause of the distancefrom the cochlearaqueduct.The data alsosuggestthat glyeinecontentcan be usedas an indicatorof CSF contaminationof ST perilymphsamples. bendand the resultingdisplacement of the tectorialmembrane.According to the experiments,the radial stiffnessof the tectorialmembraneis severaltimessmallerthanthe corresponding stiffness of the stereocilia determined by StrelioffandFlock(HearingRes.15, 19-28). Thiscontradiets the classical model of cochlear hair-cell stimulation. The tectorial membraneappearsto act on the stereociliaasa massload rather than their stiff anchor. 9:00 HH3. Three-dimensional analysisof fluid flow betweenthe teetorial membrane andtheorganof Corti.Joseph A.M. Boulet(Departmentof Engineering Science andMechanics, University ofTennessee, Knoxville, TN 37996-2030)and J. BarneyHolt (SpaceTransportation Systems Division,RockwellInternational,$$5 DiscoveryDrive, Huntsville,AL 358O6) S70 J. Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 9:30 HHS. Two-microphonemeasurementof the power-refleetioncoefficient in humanears. William J. Murphy, Arnold Tubis (Departmentof Physics,Purdue University,West Lafayette,IN 47907), and Glenis R. Long (Department of Audiology and SpeechSciences,Purdue University,WestLafayette,IN 47907) Foreachofeightnormalhearing andeightabnormal hearingsubjects, the acousticintensityand powerreflectioncoefficient spectraare measuredusinga two-microphone probewithprobetips• 8 mmdownstream from the pointof sounddelivery[W. $. Murphy, A. Tubis,and G. R. Long, J. Acoust. Soe. Am. Suppl. I 81. S75 (1987)]. The tube oran EtymoticResearch ER-2 sounddeliverysystem passes throughoneport and the probetipsfrom a pair of ER-7 microphones passthroughthe secondport of a latexear mold.Resultsin the frequencyrangeof 4-12 114thMeeting: Acoustical Society ofAmerica S70 kHz are compared with previous measurements by Stinsonet al. [J. Acoust.Soc.Am. 72,766-773(1982)], Hudde[J. Acoust.Soc.Am. 73, The characteristic equationfor a simplemodelof the externalear is usedto demonstrate theindividualeffectsof theradiationimpedance, of 242-247(1983)], andStinson [J.Acoust. Soc.Am.77,386-393( 1985); thepinna,andof theearcanal,including a wedge-shaped termination at 79, 1003-1009(1986)]. Hearingthresholds arealsodetermined in order the tympanicmembrane, on the resonance frequencies of the ear.The toidentify anysignificant correlations intheobjective andpsychophysical resultsshowthat a denotationsuchas "first quarterwaveresonance," measure. [Worksupported bya grantfromtheDeafness Research Founsuggesting a physical dimension oftheear,whichisa quarterwavelength dation.] long at the resonance frequency,is a misnomer.[Work supportedby NIH, NINCDS. ] 9:45 HH6. The relationbetweenthe impedance at the microphone and the transferimpedance of an occludedear simalator.GeorgeF. Kuhn (VibrasoundResearchCorporation, 2855 West Oxford Avenue, 10:15 Englewood,CO 80110) TheANSI $3.25-1979 Standard foranoccluded earsimulator specifiesthetransferimpedance, thatistheratioof thepressure at themeasuring microphoneto the volumevelocity12.7 mm in front of the micro- phone, from100-10000Hz. Thispaperpresents analytical relationships between thetransferimpedance andthesimulated eardrumimpedance. Limitof errorcalculations areusedto derivetheexpected tolerances for oneimpedance whenthetolerances arespecified fortheotherimpedance. [Work support• by NIH, NINCDS. ] HHS. Absorption,phase velocities,characteristlcimpedancesand transmission matricesfor tubesof smalldiameter.GeorgeF. Kuhn (Vibrasound ResearchCorporation, 2855 West Oxford Avenue, Englewood,CO 80110) Numericalresultsfor theabsorption, thephasevelocities, thecharacteristicimpedances andthetransmission matrices, relatinginputpressure andvelocityto outputpressure andvelocity,arepresented for tubeswith smallradii relativeto the boundarylayerthickness.Theseexactresults are comparedto resultsbasedon other approximateanalyticalmodels andtheirrangeof validityin termsof frequency andtuberadiusisexam- ined.Theexactanalytical modelyieldselectroacoustic analogs, forcer- 10:00 HH7. Longitudinal resonancesof the human, external ear: A first approximation.GeorgeF. Kuhn (VibrasoundResearchCorporation, 2855 WestOxfordAvenue,Englewood,CO 80110) tain sizesand frequencies, whichcontainnonrealizable negativeresistances. It isshownthata change in thelengthof thetube(s)comprising thetransmission lineovercomes thisdifficulty.[ Worksupported byHIH, NINCDS. ] 10:30-10:45 Break 10:45 HH9. A receptor-codedcochlearimp]ant: Speechdiscriminationand comprehensionresults. Gerald S. Wasserman (Department of Psychological Sciences, PurdueUniversity,W. Lafayette,IN 47907) and Richard T. Miyamoto (Department of Oto]aryngo]ogy,Indiana UniversityMedicalSchool,Indianapolis,IN 46202) Cochlearimplantsare,amongotherthings,artificialreceptors. Like haircells,theytransduce acousticenergyintobioelcctricsignalsthatactivateauditorynervefibers.Thecontribution of receptorcodingto perception in adult andjuveniledeaf patientssurgicallyfitted with cochlear implantshas thereforebeeninvestigated. A filter-feedback systemwas usedto simulatethesignalprocessing of naturalreceptors. Thissystem permittedcontrolled within-patient testsof theeffectof receptorcoding on speechperception; thiseffectwasexaminedin two differentways: Speechdiscrimination wastestedwith CVC monosyllables that differed only in their final phoneme.Speechcomprehension wastestedwith a modification of the SPIN test. All testswere run as two-alternative forced- choicetasks.Patientswerecompletelyalonein a windowless acoustic boothduringtesting; stimulation anddatacollection wereundercomputer control.In everycase,receptorcodingproduceda significant improvementin phonemediscrimination. The sentence comprehension taskwas moredifficultandmostpatientswereat or nearchancein bothexperimental andcontrolvonditions; however,receptorcodingcanproduceclearly Hearing Laboratoryof the South,ENT Department,LSUMC, New Orleans,LA 70112),andCedricF. Walker (Departmentof Biomedical Engineering, TulaneUniversity,New Orleans,LA 70118) TheNucleusWSP-III isa speech processor forcochlearimplantsthat stimulatestwo electrodes per fundamentalperiod.The electrodes stimulatedare chosenbasedon the first two formantfrequencies: lowerfrequencyformantsexciteelectrodes thataremoreapical.Thisgivesusersa cueto discriminate vowelswithdifferentformants.But,if twoparticular vowelshad secondformantscloseenoughto fall within the frequency band of some electrode, the user would miss that cue--those two vowels would "behave"asif they had exactlythe samesecondformant. Cochlear implantusersweretestedunderthreefrequency-to-electrode maps:the standardmap (usedby mostNucleusimplantpatients),the modified standardmap (proposedby P. Blarney,privatecommunication,1987) anda newmapdesignedto maximizecontrastbetweendifferentvowels.A discrimination testwasusedwheresubjects werenotaskedto identifythe soundsthey heard,but only to indicatewhetherthe differentword in a set of threewasin theinitialor finalposition.Thistendsto minimizelearning effects dueto priorexperience witha givenmap.Also,if cochlearimplant userscan discriminatedifferent phonemes,they could conceivablybe trained to identify them correctly, whereaslack of discrimination makes identification impossible. [Work supported byCochlearCorp.andSigma Xi.] significant performance in a patientwhosecontrolsentence comprehension is at chance. 11:0o 11:15 HHlO. Vowel discriminationusing different frequency-to-electrode mapsin cochlearimplantusers.Mario A. Svirsky(KresgeHearing Laboratoryof theSouth,ENT Department,LSUMC, New Orleans,LA 70112andDepartmentof Biomedical Engineering, TulaneUniversity, HHII. Evaluationof a multichannelcochlearimplant. Sigfrid D. Soli, VirginiaM. Kirby, ChristophervandenHonert,andGregoryP. Widin (Hearing ResearchLaboratory,270-4S-11,3M Center,Saint Paul, MN NcwOrlcans, LA70118),JohnK.Cu!len,HowardP.Ragland(Kresge 55144) S71 J. A,coust. Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S71 A wearablemultichannelsignalprocessor for stimulationof single- electrode cochlear implants hasbeenfieldtested withtwopatients. Each channelin theprocessor, whichisimplemented in a digitalsignalprocessingchip,consists ofa resonator followedbyaninstantaneous compressive nonlinearity.The channeloutputsaredigitallymixedfor usewith singleelectrode implants.The resonators performa spectral-to-temporal transformationof the inputsignalandthenonlinearities limit outputlevelto emulatethe responsecharacteristics of normal auditoryneurons.The resonator andnonlinearity parameters areadjustedtoaccommodate both theacousticpropertiesof speechsoundsandtheelectricaldynamicrange of the patient.Severalprocessor configurations with differentresonator andnonlinearity designs havebeenevaluated. The resultsof psychophysicaltests,usedto fit eachprocessor configuration to thepatientandmeasurespeech performance in quietandnoisewith eachconfiguration, will be reported. taneousmaskingphenomena to VllIth nerveor higherneuralmechanisms.[Work supportedby NIH.] 11:45 HH13. Chinreel interactions measured by forward-masked "place" tuning curves with multichannel electrical stimulation. Virginia M. Kirby (Hearing ResearchLaboratory,270-4S-11,3M Center,Saint Paul, MN 55144), David A. Nelson (Hearing ResearchLaboratory, Universityof Minnesota,Minneapolis,MN 55455), Sigfrid D. Soli (Hearing ResearchLaboratory,270-4S-I1, 3M Center,SaintPaul, MN 55144), and Todd W. Fortune (Hearing Research Laboratory, Universityof Minnesota,Minneapolis,MN 55455) Simultaneous 11:30 HHI2. Backward and forward maskingfor direct electrical stimulation of the VIIIth nervein two profoundlydeaf subjects.L. J. Dent and B. S. Townshand(Stanford ElectronicsLaboratories,Stanford,CA94305) Two profoundlydeafmultielectrodeimplantsubjectswererequiredto detecta probesignal( 10 msin duration) in a temporalgapbetweentwo pulse-trainmaskers(each300msin duration).The detectionthreshold wasmeasuredfor a probecenteredtemporallyin the gap,aswell asfor a probeoffsetfrom centerby up to 97.5%. Also presentedwerethe pure backwardand pureforwardmaskingcases.Qualitatively,bothsubject's forward and backwardmaskingfunctionsapproximatedthoseobserved for normalhearingsubjects[L.L. Elliott,J. Acoust.Soc.Am. 34, 11161117 (1962)] in that forward maskingdecayedmore graduallythan backwardmaskingas a functionof probe-maskerseparation.Because mechanical(cochlear)contributions to masking[H. Duifhuis,J. Acoust. stimulation of multichannel intracochlear electrodes cangiveriscto peripheralandcentralchannelinteractions. The abilityto eliminateor predictand controlthe interactionsproducedby a given electrode geometry isa processing goalforoptimizingperformance witha multichannelcochlearimplant.Previousstudieshaveusedloudness summationand forward-maskingpatterntechniquesto estimateinteractions betweenchannelsof electricalstimulation.In this study,interactionsbe- tweenbipolarchannelsof analogelectricalstimulationwereestimated using a forward-maskingparadigmwith a fixed-level,fixed-location probe.By varyingthe electrodelocationof a 200-Hz, 300-mssinusoidal maskeranddeterminingthelevelof themaskerat eachlocationnecessary to just maska 200-Hz, 10-msprobe,a "place"tuningcurvewasderived. The levelofmaskerrequiredat a givenlocationto maskthe probedepends Soc. Am. 54, 1471-1488 (1973) ] can be excludedin the caseof direct on the amountof excitationproducedby the probeandreflects,in part, the degreeto whichthereis overlapof neuralpopulations responding to eachstimulus. These"place"tuningcurves,whichdisplayinteractions as a functionofmaskerlocationweredeterminedfor severalprobelevelsand probelocations.Resultsand implications for speechprocessing strategies VIIIth nervestimulation,thesedatasupporttheattributionof nonsimul- will be discussed. THURSDAY MORNING, 19 NOVEMBER UMS 4 AND 5, 8:30 A.M. TO 12:00 NOON 1987 SessionII. UnderwaterAcousticsVI. SignalProcessingfor UnderwaterAcoustics(Prrcis-PosterSession} Thomas G. Muir, Chairman NATO S•4CL•NT Centre,La Spezia,Italy Chairman's introduction---8:30 ContributedPapers Followingpresentation of the prrcis,posterswill be on displayuntil 12:00Noon. 8:35 111. Nonorthogonality of measured normal modes in shallow water. Grayson H. Rayborn (Naval Ocean Research and Development Activity, NSTL, MS 39529-5004 and Department of Physics and Astronomy,Universityof SouthernMississippi,Hattiesburg,MS 39406), GeorgeE. Ioup, and JulietteW. Ioup (NORDA, NSTL, MS 39529 and Department of Physics and Geophysical Research Laboratory, University of New Orleans,New Orleans,LA 70148) The importance of surmounting the nonorthogonality of measured normalmodesandprocessing shallow-water datain sucha waythatmod$72 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 al compositions areeffectively recovered for matchedfieldprocessing has beendemonstrated by severalinvestigators. The potentialfor improvementusingthistechnique is greatest whenthe nonorthogonality of the measured modesislargest.Theamountbywhichthenormalmodesfail to beorthogonal fora varietyofoceanbottoms, arraylengths anddiseretizations,andarraypositions for the Pekerismodelhasbeenstudied.Environmentsare selectedto reflectsedimenttypescharacteristicof the continentalshelf.It hasbeenfoundthat the nonorthogonality is greatestfor waterdepthsandfrequencies at andjustabovemodalonsets, for soundspeedratioscloseto 1, and for arrayswhichspanonly a fractionof the watercolumn.Judicious placement andlengthselection for shortarrays, however, cangiveorthogonal measured modes forsomecombinations of frequencies andenvironments whena smallnumberofmodes arepresent. 114thMeeting:AcousticalSocietyof America $72 8:39 I12. Effects of measuredmode nonorthogonalityon conventional matchedfield proeessing. JulietteW. Ioup, GeorgeE. Ioup (Naval OceanResearchand DevelopmentActivity, NSTL, MS 39529-5004and Departmentof Physics,Universityof New Orleans,New Orleans,LA 7014g), Grayson H. Rayborn (NORDA, NSTL, MS 39529 and Department of Physics and Astronomy, University of Southern Mississippi, Hattiesburg,MS 39406), DonaldR. Dei Balzo(NORDA, NSTL, MS 39529), andChristopherFeuillade(ODSI DefenseSystems, peaks.Thispossibility of loweringtherequirements onourknowledge of theenvironment is investigated with two methods:( 1) by constructing multiplebeam(constraint)algorithms and(2) byconsidering stochastic blurringbythemedium.Thesetwoapproaches areappliedto plane-wave beamforming andmatchedfieldprocessing. 8:51 Inc., 6110 Executive Boulevard, Suite 320, Rockville, MD 20852) As discussed by Raybornet al. {precedingabstract),the potential improvementachievable from the useof modalfilteringin matchedfield processing is greatest whenthemeasured modesareleastorthogonal. To assess the abilityof matchedfieldprocessing usinga conventional crosscorrelationestimatorto realize this potential,ambiguitysurfacesfor a Pekeriswaveguidehave beenconstructedemploying(a) the pressure fieldsat the hydrophones and (b) the amplitudesresultingfrom modal filtering.Thequalityof thesurfaces isquantified utilizingmeasures that comparetheheightof thesourcepeakabovethemeanandtheheightof the sourcepeakabovethe standarddeviationof the surface.The results indicatethat modalfilter processing, for a givenarray of hydrophones, offersthe mostimprovementwhen bottomdepth,bottom type, and frequencycombineto producemeasured modesthat arethe leastorthogo- nal,andyieldsno improvement whenthemeasured modesareorthogonal. IIS. Spatial matched processing for multipath propagation. Matthew Dzieciuchand T. G. Birdsall (Communieationand Signal ProcessingLaboratory,4242 œœCSBuilding, Departmentof Electrical Engineeringand ComputerScience,The Universityof Michigan,Ann Arbor, MI 48104) Underwateracousticpropagationis characterizedby multipathor multimodepropagation.Ray theoryand modethenryare not fully adequatefor modelingphysicalreality.Impulseresponses canbemoreaccuratelycalculatedusingGaussianbeamtheory.Signalprocessors can be designed to takeadvantage of thechannelcomplexity if thepropagation is actuallyknownsothatdetectability isincreased. The proposed technique, channelmatchedfiltering,synthetically backpropagates thewavefrontto a hypothesized sourcelocation.Accuratepassive estimates of sourcelocation can be madewithout knowledgeof the signalcharacteristics. GB theorycan easilyaccommodate a range-dependent deepwaterenvironment. [This researchsupportedby theOfficeof Naval Research. ] 8:43 II3. Effects of correlatednoiseon the cross-spectralmatrix in modal 8:55 composition space.GeorgeB. Smith, ChristopherFeuillade (ODSI Defense Systems,Inc., 6110 Executive Boulevard, Rockville, MD 20852), and Donald R. Del Balzo (Naval Ocean Research and Development Activity,Code244,NSTL, MS 39529-5004) Computersimulationsof hydrophonecross-spectral matricesin a shallow-waterwaveguideweregeneratedfor signalswith differentmixturesof,correlatedandwhitenoise.Thesematriceswerethenmappedto modalcomposition space,a spacepopulatedby vectorswhoseelements are the amplitudesfor the trappedmodes.It wasfoundthat, in modal compositionspace,the cross-spectral matrix is not sensitiveto the differencebetweencorrelated and white noise,but is sensitiveto the difference betweennoiseandsignal.While thedistributionof signalandwhitenoise amongtheelements of thecross-spectral matrixissimilarbeforeandafter the mappingto modalcomposition space,the distributionof correlated andwhitenoiseisnot.Temporallydiscretenoisesources, whicharecorrelatedat thehydrophones, butnotfromsampleto sample,makelittleor no contributionto the off-diagonalelementsof the cross-spectral matrix in modal compositionspace.This fact has significantimplicationsfor matchedfieldprocessing in low signal-to-noise situations. lI&. A new techniqueof acousticmodefilteringin shallowsea.Harish M. Chouhan and G. V. Artand (Department of Electrical Communication Engineering, IndianInstituteof Science, Bangalore 560 012, India) A newtechniqueof filteringacousticnormalmodes,whichovercomes manyofthedrawbacks oftheearliertechniques, ispresented in thispaper. It is basedon the fact that eachnormalmodein an isovelocitychannel comprises a pairofplanewaveswithcharacteristic directions of propagationsymmetrically disposed with respectto thechannelaxis.A vertical array of equispaced hydrophones is shadedso as to steer nullsin the directionalresponse of the arrayalongthe directions of arrivalof plane wavescorresponding to theunwantedmodes.All theshadingcoefficients arereal,leadingto simplicityin thehardwarerealization of theprocessor. Theshading coefficients areinvarianttoa shiftin thepositionofthearray. Thearrayneednotspantheentiredepthof theocean.Efficient filteringis possible evenwhenthe eigenfunctions of the modeshavea significant penetration into thebottom.[Work supported by DOE, Government of India. ] 8.-47 8:59 I14. Adaptive beamformingor matchedfield processingin media with uncertain propagation conditions. A. B. Baggeroer, H. Schmidt (Massachusetts Institute of Technology,Cambridge,MA 02139), W. 117.Matched eatastrophealeconvolution with applicationto the inversion ofmarineseismicrefraefiondata. Michael G. Brown (RosenstielSchool A. Kuperman(Naval ResearchLaboratory,Washington,DC 20375}, andE. K. Scheer(WoodsHole Oceanographic Institute,WoodsHole, of Marine and AtmosphericScience,University of Miami, 4600 RickeobackerCauseway,Miami, FL 33149) and Paul E. Bullwinkel MA 02543 ) (Applied MeasurementSystems,Inc., 1415 S.W. 21st Avenue, Fort Adaptivebeamforming or matchedfieldprocessing provides highresolutionwith sidelobe controlif accuratereplicafieldscanb• generated. The generation of thesereplicafieldsis a formidable problemrequiring knowledgeof the complexoceanpropagationenvironment.On the other hand,thedetectionproblemin theoceanmaynotrequirehighresolution, whereassidelobecontrol is still an important issue.Relaxingresolution requir•nentssuggest that a certaintoleranceincorporatinguncertaintyof the propagationconditionsispermissible,or evendesirable,becauseof the difficulties bothin specifying themediumexactlyandin identifying global $73 J. Acoust.Soc.Am.Suppl.1, VoL82, Fall 1987 Lauderdale, FL 33312) The problemof extractingaccurateestimatesof the travel timesof unresolved arrivals(e.g., the secondand third arrivalswithin a triplication ) from a set of noisy bandlimited measurementsof a time-dependent acousticwavefieldisaddressed. The methodof solutionpresented isbased on the assumption that the underlyingcausticstructure(or, equivalently, the traveltime curvestructure)of the wavefieldis known.Because genetic causticsassociated with causalwavefieldstakeon only certainforms, thisis a weakassumption. Additionally,it is assumed that the medium 114thMeeting:Acoustical Societyof America S73 structurevariesasa functionof depthonly.Thisfurtherrestrictstheclass improvedby usingvariousprefiltering techniques to process theexperi- of observablecausticsections.Theseconstraintsare incorporatedinto a mentaldata.Eachof the prefilters--Hannan-Thomson, smoothed coher~ encetransform,and phasetransform---consists of a frequency-domain weightingfunctionthat is appliedto thecross-spectral densityfunction beforethecross-correlation functionis computedusingthe inverseFourier transform.The prefiitersare observedto enhancethe estimationof the timedelayby reducingtheambiguityassociated withdetectingthepeak valueof thebasic(unweighted)cross-correlation function,whichisoscil!story in nature. matchedfield processing approachto the deconvolution problem.The techniqueis demonstratedusinga measuredmarineseismicrefraction dataset.A Herglotz-Wiechertinversionof the resultingkinematicdata T(R ) agreeswell with previouslypublishedwaveforminversionsof the samedataset.Advantages of thenewapproach overlinearizedwaveform inversions are discussed. 9:03 9:15 IIg. Effects of nearby soft screens on idealized array reception. Rudolph Martinez and Ann Westagard (Cambridge Acoustical Associates,Inc., 54 CambridgeParkDrive, Cambridge,MA 02140) Arraysof sensors mustoftenfunctionin theneighborhood of riffratting and reflectingsurfaces;these,in turn, representeffectivepassive sourcesthat add to the incomingsignal.The resultis a computedwavenumberspectrumcontaminatedby the presenceof spurioustargetsin whichthe true seasourcemay appearshiftedoff itsactualazimuth.Here, sucheffectsare demonstratedfor the fully analyzablesituationof an unshadedarray neara thin semi-infinitesoftscreen,takeneither asa colinear aft baffleor a perpendicularbackstopto the incidentwave. Exact and asymptotic expressions havebeenderivedto predictperceived strengths of virtualtargetsandassociated azimuthaldistorsions; theirreportednumericalevaluations couldserveasbenchmarkresultsagainstwhichone mightgaugetheeffectiveness of sophisticated processing methodsunder similar array installationconditions. 9.'O7 II9. Limitationsto sourcedescriptionset by a finite recei 'ing aperture. Philip L. Stocklin (Consultingphysicist,439 Blue Jay Lane, Satellite Beach,FL 32937}, and Norma L. Stocklin(SystemsAnalyst, 439 Blue Jay Lane, SatelliteBeach,FL 32937} II11. SteppedFM signalsfor acousticremote sensingof geoaeoustic properties of bottom sediments. Harry A. DeFerrari and Hien B. Nguyen (Department of Applied Marine Physics,R.S.M.A.S., Universityof Miami, Miami, FL 33149) Signalsusedfor sub-bottomprofilingare usuallyhigh-intensitytransientsgenerated bycxplosious, air guns,sparkers, etc.Eachtransienthasa uniquecomplexfrequencyspectrumand serialtransmissions cannotbe coherentlyaveraged.Coherentacousticsignalsgeneratedby electromechanicaltransducers arenot widelyused,owingto the relativelylow intensityor bandwidthlimitations,but evenweaksignalscan be averaged for longperiodsif thepropagation mediumisstable,to produceveryhigh equivalentsignallevels.Likewise,it appearspossibleto synthesizevery broadband signals by coherently averagingnarrow-band signals.Several narrow-bandsignalstransmittedseriallycanbeprocessed asthoughthey were transmittedsimultaneouslyto synthesizesignalsof any desired bandwidth--amethodcalledsteppedFM. A systemhasbeendeveloped andtestedfor transmittingandreceivingpseudorandom codesteppedFM signals.Signalswith a bandwidthof 1700Hz and intensitylevelsgreater than 2 i0 dB after processing are usedto probethe firstfew tensof meters of thebottomsediments. The resultingpulseduration,approximately 0.5 ms,allowsfor a separation of reflectedfromrefractedpaths.Traveltimes of refractedrays can then be usedfor tomographicinversionsto yield compressional velocityprofilesof the bottomsediments. Spatialsamplingfunctiontheoryresultsare statedfor a monochromatic acoustic field and for a bandwidth-limited resultsare appliedto a finitereceivingapertureto estimatethe physical limitationsto sourcedescriptions obtainable fromtheoutputsofelements of the aperture.Effectsof variationsin propagationconditions,received signal bandwidth,receivedSNR, and available processingtime are shown.Relationships to inversefieldprocessing are developed. 9:11 II10. Improvedtime delayestimatesof underwateracousticsignalsUS/he beamforming andprefilteringtechniques. BrianG. Ferguson(Weapons Systems Research Laboratory, Defence Science and Technology Organization;Royal AustralianNavy ResearchLaboratory,P.O. Box 706, Darlinghurst2010, Australia) The cross-correlation methodfor estimatingthe differentialtime delay for an underwateracousticsignalto arrive at two spatJellyseparated receiversis described.The output amplitudesof the two receiversare crosscorrelatedand the lag time at which the cross-correlation function peaksprovidesan estimateof the time delay.Often the time delayestimatesof thesignalare corruptedby the presence of noise.By replacing each of the (omnidirectional) receiverswith an array of receiversand thencrosscorrelatingthebeamformed outputsof thearrays,it is shown thattheeffectof noiseonthetimedelayestimationprocess issubstantially reduced.Bothconventional and adaptivebeamformingmethodsare implementedand theadvantages of arraybeamforming(prior to crosscorrelation)arehighlighted usingbothmodeleddata( fortutorialpurposes) andexperimental data.The performance of thecrosscotrelatoris further S74 9:19 acoustic field. These J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 1112. Signal design for Doppler processing. Ziad Haddad and BowenParkins(AT&T Bell Laboratories,14A420,WhippanyRoad, Whippany,NJ 07981) The Dopplersonarprocessing requires that theilluminaJting signal havea spectrumwithlow sidelobes overtherangeof frequencies at which Dopplershiftsare expected.This is conventionallyachievedby "windowing" the driving sinewave.Uncertaintiesin the amplituderesponse (thresholdonsetandlinearity) of typicaltransducers makeit desirableto havealternatemeansof obtaininglowsidelobes. Oneapproachisto transmit a seriesofnonuniformlyspacedshortpulsesof constantdurationand amplitude.Pulsestart timesare specifiedto yield the desiredspectral shapeusingtheapproachofA. Ishimaru[ IEEE Trans.Ant. Propag.AP10, 691-702 (1962)] developedfor spacetaperingantennaarray elements.The resultingspectrumis not affectedby nonzerothresholdonset or limitedlinearrange.The sidelobereductionthat can be achieved pendsonthenumberof pulses,hence,ontheamplifierbandwidth.Amplitude shadingand varying pulsedurationsallow the de.signofa sec. ond typ• of signal,a windowedcw signalthat doesnot dependon thresholdonset and is less sensitiveto amplifier bandwidth. The sidelobereduction achievablewith this seconddesigndependson the linear range. 9:23 II13. Phaserecoveryand calibrationwith underwateracousticarrays. Nathan Cohen(MetropolitanCollege,BostonUniversity,Boston,MA 02115) Underwateracousticimagingprovidesuniquechallenges because of the nature of the medium and difficulties in accurate measurement of 114thMeeting: Acoustical Society ofAmerica S74 arrayobservables. Baseline uncertainties andlossof phaselockingmake phaseobservables especially difficulttoacquireinsomecases. Here,linear and nonlinearimagingtechniques that.aid in thecalibrationand/or recoveryof phaseobservables andtheself-calibration ofamplitudeobservablesaredescribed. Enhanced dynamicrangeandimageaccuracy arepossible with these post-processing techniques,which include closure relationsand phasereferencing. Examplesof theapplicationof the techniquesin otherimagingdisciplines, suchasopticsand radio astronomy, are discussed. 9:27 II14. Interpretation of coherenceestimatesdetermined for time seriesof limited length. Allen E. Leybournc,III (Department of Engineering Technology,University of SouthernMississippi,Box 5172 S. Station, Hattiesburg,MS 39406-5172) the calculationof the reverberation andnoisebackgroundalongwith the signalpropagation characteristics. To consider thegeneralcaseofa threedimensional beampatternfor bothsourceandreceiver,andthreedimensionalnoisefields,thepredictionmodeltendsto beexpensive in computer time and memory requirements.The ASONAR model presentedhere incorporates theRUMBLE bistaticreverberation model[ H. P. Bucker,J. Acoust.Soc.Am. Supp.1 80, S 63 (1986) ] and the DUNES directional noisemodel[ R. W. Bannister,A. S.Burgess, andD. J. Kewley,J. Acoust. Soc.Am. Suppl.I 80, S65 (1986) ]. Thesenewlydeveloped modelsincorporatefast executionspeedsand can run on IBM PC type microcomputers.By usingSALT (_Sound-A_ ngle-_Level-Time) tables,the propagation lossand reverberationto the many scatteringareasare quickly determined.The scatteringareasare increasedin sizeas the rangesincreaseto keepthe computationtime within reasonable limits.The noise calculationsare madeusingsimplifiedpropagationlaws.ASONAR combinesthe outputsof RUMBLE and DUNES to calculatebistaticsignal excessversusrangefrom the receiver.Comparisonswith other models andsomeexperimental dataaremade.[ Work supportedby theNORDA NOP Programand NUSC. ] Coherenceof multichanneltime seriesdatahasbeencalculatedby the methodof G. C. CarterandJ.F. Ferrie[A Coherence andCross Spectrum EstimationProgram,Programsfor Digital SignalProcessing ( IEEEoNew York•1979),No. 2.3-1-18]. The properties of thisstatistichasbeencontrastedthroughsimulationstudiesat known noiselevelsfor shortand longtime seriesthrougha rangeof 40 datasegments. Resultsat lessthan eightdatasegments are shownto bedifficultto interpret;however,modestimprovementsin the coherenceestimatecanbe madeby adjustments indicatedby thesestudies.A methodof calculatingcoherence estimates througha techniquereferredto as frequencybin averagingis described. The results obtained are contrasted with the conventional method of Carter. Although reliableestimationof coherencefor short time series data (transient data) is difficult to obtain in the conventional sense,the comparisons madeshouldprovidea basisfor interpretationof coherence estimates aswell assuggest corrections that may be appliedto relatively shorttime seriessequences. [Work supportedby NORDA.] 9:31 II15. The effect of multipath and sensor location offsets on the performanceof a cross-correlation system.P. Bilazarian (Raytheon Company, Submarine Signal Division, 1847 West Main Road, 9:39 II17. The designand performanceanalysisof barrel staveprojectors.D. F. Jones and G. W. McMahon (Defence Research Establishment Atlantic, P.O. Box 1012, Dartmouth, Nova Scotia B2Y 3Z7, Canada) There is growing interest in low-frequency,light-weight sound sourcesfor underwateracoustics researchandfor potentialfuturesonar applications. A designbeingstudiedat DREA is the barrelstaveprojector, which consistsof a piezoceramic cylinderdrivinga set of annular staves in flexure. An electroelastic finite element model has been used to analyzethelinearvibrationsandacousticradiationsof projectorshaving outer diametersof about 8 cm and fundamentalresonance frequencies below 1600Hz in water. Prototypeprojectorsthat operateat electroacousticefficiencies greaterthan 50% at resonancehavebeenassembled andtheirmeasured acoustic properties havebeencompared to thosepredictedby the model.In addition,a brief discussion of the modelparametersand limitationsis given. Portsmouth, RI 02871 ) The combinedinfluenceof multipleray arrivalsandsensorlocation offsetson the performanceof a broadbandcross-correlation systemis investigated. Thesystem consists of threesensors thatarenominallypositionedon a horizontalline. Sourcelocationparametersare determined from time-delayestimates obtainedby cross-correlation techniques. An analyticalmodelfor the predictionof localizationbiaserrorsdueto multipathandto offsetsfrom nominalsensorpositionsisdiscussed. This model is applicableto an oceanicmediumwith a depth-dependent sound-speed profile.Resultsarepresented for a varietyofmultipathconditions, suchas thoseassociated with directpath,surfaceduct,bottombounce,and convergencezonetypesof propagation.The sensitivityof localizationerrors to variationsin environmentalparameters,suchas seastateor bottomreflectionproperties,is also investigated.It is demonstratedthat multipath, especiallywhencombinedwith sensorpositionoffsets,can havea significantimpacton systemperformance. 9:35 II16. A fast bistatlc reverberationand systemsmodel. D. J. Kewley (Naval Underwater Systems Center, New London, CT 06320 and WeaponsSystems ResearchLaboratory,DSTO, Departmentof Defense, GPO Box 2151, Adelaide, SA, Australia), and H. P. Bucker (Naval OceanSystemsCenter,SanDiego, CA 92152) The performance predictionof bistaticactivesonarsystemsrequires $75 J. Acoust.$oc. Am. Suppl. 1, Vol. 82, Fall 1987 9:43 I118. A practical VME-based sonarsystemdevelopmentworkstation. Shon Sloat(AP Labs,4411MorenaBoulevard,Suite150,SanDiego,CA 92117) Thispaperdescribes presentcapabilityandfurturedevelopment plans of a practicallow-costsonarsystem-development work stationbasedon off-the-shelf VME equipment. The workstationallowssystemengineers and algorithmdesigners to quicklydevelopsonarprocessing algorithms andsystems. The workstationcanbeusedto (a) simulateandanalyzethe characteristics of realtimeenvironments for systemandalgorithmdevelopment,(b) concurrentlydevelopfinal implementation software,and (c) generate analogtestdataforsystem testandoperatorinstruction. The work stationovercomesmany of the difficultiesexperiencedwith developingsonaracousticsignalacquisitionand processingsystems:high designandmaintenance costsof customequipment;lackof controllable testdata;inabilityto correlatesimulationdata with final systemperformance;andhighsoftwaredevelopment costs,particularlywith parallel, pipelinedsignalprocessing equipment.The sonardevelopment work station hardwareconsistsof a 16.67-MHz 68020 processorsupportedby a VME busarchitecture.The hardwareprovides(a) internrivecolorgraphics, (b) Sky Warrior and Mecury Zip 3232 array processors. The work stationsoftwareis basedon the Unix development system,VRTX real time kernel,X-Windowsstandardcolorgraphicssoftware,andarray processorsonarprocessing primitives.This foundationprovidessoftware portabilityand deviceindependentgraphics. 114th Meeting:AcousticalSocietyof America S75 THURSDAY MORNING, 19 NOVEMBER 1987 TUTTLE CENTER ROOM, 9:00 A.M. TO 12:05 P.M. SessionJJ. Physical AcousticsV: Acousticand SeismicCouplingI: Air-Solid Interfaces Gilles A. Daigle, Chairman Divisionof Physics,National ResearchCouncil,Ottawa, OntarioK1,4 OR6, Canada Cboirrnan's Introductions9:00 Invited Papers 9:05 JJ1. Near-groundsoundfieldsandsurfacesof finite impedsnee. TonyF. W. EmbletonandGillesA. Daigle (Divisionof Physics,NationalResearchCouncil,Ottawa,OntarioK 1A 0R6, Canada) Studiesoverthe past30 yearshaveelucidatedthe effectsand interactionsof manymechanisms involvedin soundpropagation outdoors.Interactions with thegroundsurfaceareespecially importantfor manyapplicationsto practicalproblems.It isparticularlysignificantthat groundsurfaces areporous,and thushavea finite complexacousticimpedance. The historyof soundfieldmeasurements to deducevaluesof impedance, andits directmeasurement, is reviewed.Suchdata prompted,and werethen usedto validate,modelsfor ground impedance, firstin termsof a singleparameter(flowresistivity of thegroundsurface)andlaterin termsof one to threeadditionalparameters. Impedancevaluesfor manygroundsat low frequencies aredifficultto measure accuratelyand severalnewtechniques havebeendeveloped.Nonisotropicandnonhomogeneous groundhas beenmodeledasa layeredmediumor asa fluid-filledporousmatrix. Theseeffortshaveincreasedour understandingof the acousticalpropertiesof the groundasan air-solidinterfacefrom a purelyempiricalbasisto a detailedphysicaldescription basedon themicroscopic properties of theground. 9:35 JJ2. Airborne soundto seismiccoupling:Backgroundfor developmentof the models.JamesM. Sabatier, Henry E. Bass,and Lee N. Bolen(PhysicalAcousticsResearchLaboratory,The Universityof Mississippi, University,MS 38677) The initial measurements of the acoustic-to-seismic couplingphenomenausingloud speakersas sound sources weremadeby the Universityof Mississippi andWaterwaysExperimentStation.For thesemeasurements,a speakerwassuspended froma crane,thesoundlevelin thefarfieldwasmeasured with a microphone abovethesurface,andthegeophone response wasmeasured belowthesurface.The resultsof thisworklargely indicatedthat the geophone hada response 1000timesgreaterthanonewouldexpectfromcalculations based uponsimpleacoustictransmission througha boundarybetweentwo perfectfluids.Anotherfindingwasthat thetransittimesfromspeakerto microphone andspeakerto geophone wereapproximatelythespeedof sound in theair. Success in ourlabwithwater-saturated sediments suggested theapplication of theBlotmodelto the ground.Microphonesweredeveloped that wereusedasa porefluidprobein the soil,andthe attenuationand phasespeedin the pore fluid were measured.This work led to the understandingof the commonlyassumed locallyreactingpropertyof the ground. 10:05 JJ3. Airbornesoundto seismiccoupling:Presentunderstanding. K. Attenborough(EngineeringMechanics Discipline,The OpenUniversity,Milton KeynesMK7 6AA, UnitedKingdom) andJ. M. Sabatier(Physical AcousticsResearchLaboratory,Universityof Mississippi,University,MS 38677) Theoriesof outdoorsoundpropagation at near-gra•ingincidenceon porousgroundsurfaces assumeeither that the groundsurfaceis an impedanceboundaryor that it is one of extendedreaction.Suchtheoriesare consistent with modelsof theacousticbehaviorof the groundasa locallyreactingfluidor asa semi-infinite or layeredmodifiedfluid supportinga singletype of compressionalwave motion. However, thesemodelsare unableto explainthe responseof buriedgeophones to airbornesoundsources.Measurementsof the ratio of a shallowburied geophoneoutput to the output of a verticallyseparatedmicrophoneplacednear the ground surfacehavebeenexplainedin a varietyof soil typesby meansof a layeredporoelasticgroundmodel.The layeredporoelasticgroundmodeland its experimentalvalidationare described,and its implicationsfor the impedance-frequency characteristics of outdoorgroundsurfaces are explored. S76 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America S76 ContributedPapers 10•$ JJ4. Sefmnic,acoustic,andionospheric wavekinematicsassociated with moderateto large earthquakes,S. I. Warshaw, F. E. Followill (LawrenceLivermoreNational Laboratory,Livermore,CA 94550), J. M. Mills, Jr. (SOHIO Production,Dallas, TX 75221), and P. R. Albec (Radio Sciences and Instrumentation,SantaFe, NM 87501) UnusuallydetailedDopplerradarrecords ofionospheric disturbances followingtwo moderate-to-large dip-slipearthquakeswere reportedin 1984.Theseindicatethat earthquakes can perturbthe ionosphere by meansof acousticwaveslaunchedby the movinggroundsurface.For one arrivals,sincethey travel at the higherseismicwavevelocity,but their amplitudesareaboutthreeordersof magnitudelessthanthemotioncou- pledviathelater-arriving air wave.A comparison of thesummerandthe winterrecordings showstwomajoreffects of a 25-cm-thick snowcover. Frequencies above100Hz arestronglyattenuated,with signalsrecorded by geephones aboveandbelowthe snowyieldinga valueof around2 for the Q of snowat 30 Hz; the snowproducesa strongwaveguideeffectthat enhances theair-coupled Rayleighwaves.[Work supported by Directorateof ResearchandDevelopment,U.S. Army corpsof Engineers Project 4A762730AT42. ] earthquake {magnitude 6.5,Coalinga, CA, 2 May1983[I. H. Wolcottet aL,J. Gcophys. Res.89, 6835-6839 (1984)]},a seismic-aeroacousticionospheric kinematicwaveanalysis wasperformed to explainradarsignaturetiming.SurfaceRayleighwavesweremodeledand interpolated frommainandafter-shockseismicrecords,acousticwavetrajectories by acousticray tracingin a standardatmosphere wereestablished, andradar detectionaltitudesweredeterminedfrom radio rays tracedthroughan ionosphere calculatedfroman ionogrammeasured closeby.Our calculated resultsagreewellwith Dopplersignature timesobserved onseveral5and 10-Ml-Izradarbeampaths160-300km fromtheepicenter. Similari- tiesanddifferences wereidentified in theotherearthquake {magnitude 7.1,Urakawa-Oki, Japan,21March1982[T. TanakaetaL,1.Arm.Terr. Phys.46, 233-245(1984)]} andits signatures werestudied to assess commonalities in perturbationmechanisms. It is tentativelyproposed thatnonlinearacoustic propagation effects couldbesignificant in shaping observed Dopplersignatures.[Work performedunderthe auspices of USDOE by LLNL undercontractW-7405-Eng-48. ] 11:20 JJ7. Applicationof ray theory to propagationof low-frequencynoise from wind turbines. James A. Hawkins and David T. Blackstock ( AppliedResearch Laboratories, TheUniversityofTexasat Austin,P.O. Box 8029, Austin, TX 78713-8029) Raytheoryhasbeenusedtoattempttoexplainthepropagation ofvery low-frequency ( 1-20 Hz) noisegenerated by downwindwindturbines. Two NASA fieldexperiments with a largedownwindmachineat MedicineBow,Wyoming,showthatthedownwind sounddecays by spherical spreading nearthesource, butbycylindrical spreading downrange. Ray theorycalculations explainthisbehavior.The favorablewindgradient downwindcauses raysto refractdownwardandbouncealongtheground. A soundchannelis built up by repeated jumpsin the numberof ray arrivalsat the receiverasdownwindrangeisincreased. Near the source, wherenomultiplyreflected raysarrive,thesoundfieldspreads spherical10:50 JJS. Acoustic-to-seismic coupling under winter conditions. Lindarnae Peck (USACRREL, 72 Lyme Road, Hanover, NH 037551290) ly. The firstjump (onsetof multiplearrivals)is predictedto occurat about2 km for the conditionsof the NASA experiments. The second NASA experiment confirmsthis prediction.For the upwindsound,a shadowzoneis predictedbecause of theunfavorable windgradient.The NASA measurements showonly that the upwindSPL in the predicted shadowzoneisof order7-! 5 dB lessthan thedownwindSPL at compara- A fieldprogramto investigate acoustic-to-seismic couplingunder winterconditionstook placein Maine duringDecember1985-January bleranges.[Work supported by NASA. ] 1986.The testareawaspackedsandwith a top layer (30 cm) of loose sand.Site conditionswere: 20-cm frost with no snowcover, 6-cm snow or 18-cmsnow;and45-cmfrost,28-cmsnow.Microphoneswereat heights of 1/2-2 m or at the sand{snow)surface.Geephones wereplacedat the snowsurface,the sandsurface,and severaldepthsin the sandlayer. Acoustic-to-seismic couplingthroughfrozensandwasinvestigated under controlledconditions usinggeephones in a sandbed at the USACRREL FrostEffectsResearchFacility.Testconditionswere:dry sand,frozenor onfrozen;saturatedsand,frozen or unfrozen;and thawingsand.The acousticsourcefor both investigations wasblank pistolfire. Resultson acoustic-to-seismic couplingthrougha snowlayerand/orfrozensandare presented andcontrasted withsummerconditions for thefrequency hand 5-500 Hz. [Work supported by Directorateof Researchand Development,U.S. Army Corpsof EngineersProject4A762730AT42.] 11:05 JJ6. Low-frequencyaconstic-to-aeismie couplingin the summerand winter. Donald G. Albert (USACRREL, 72 Lyme Road, Hanover,NH 03755-1290) Experiments conducted in northernVermontinvestigated acousticto-seismic couplingin the5- to 500-Hzfrequencybandfor rangesbetween I and300m.Thestrongest couplingintothegroundoccursastheair wave passes, witha measured ratioof about0.001cms- •fPa.TheP waves are inducedinto the groundimmediatelyunderthe •ouree,and are the first S77 J. Acoust. Soc. Am. Suppl. 1, Vol. 82, Fall 1987 11:35 JJg. Bcam-to-modeconversionof a high-frequencyGaussianP-wave inputin anelasticplateembedded in vacuum.I.T. Lu, L.B. Felsen,and J.M. Klosner(Polytechnic University,Farmingdale, NY 11735) Gaussianbeams,either individuallyor as synthesizing basisfields, provide useful models fortheresponse duetoGaussian ormoregeneraliy shapedhigh-frequency sourceinputs,respectively. Whenpropagating intoa layeredenvironment, an initiallyweli-collimated beamundergoes diffusionafter successive reflections,and is convertedessentiallyinto the oscillatory patternofoneor moreguidedmodes. Thediffusion process is accelerated in thepresence of multiwave coupling at interfaces, forexample,theP-SVcoupling atthefreeboundary ofanelastic solid.Thisphenomenology isexamined hereforatwo-dimensional P-wave beaminputinto an elasticplateembedded in vacuum.Thebeamismodeled viathecomplexsource pointmethodwhereby a Gaussian-waisted inputisgenerated froma lineforcingfunctionby assigning complexvaluesto the source coordinates. This schemeavoidstheneedfor plane-wave spectraldecem- position andsynthesis, buttheanalyticcontinuation to complex source coordinates of variouslinc forceGreen'sfunctionrepresentations in a layeredmediumrequiresprecautions, whichare discussed. Numerical calculations of P andSg'potentials, stresses andmotions,performed by beamtrackingandmodesummation at successive crosssections, reveal the beam-to-mode transitionprocess, and suggest that a hybridbeammodealgorithmprovides the mostdirectandcogentphysicalexplanation. [Work supportedby AFOSR andONR. ] 114th Meeting: AcousticalSociety of America 877 nearthestagnation pointofa bluffbodywill beexamined. The stabilityof 11:50 J J9. Acoustic wave interactions at a mean-flow st•guation point. CharlesThompsonand Martin Manley (Departmentof Electrical Engineering, Universityof Lowell,OneUniversityAvenue,Lowell,MA 01854) suchflowswill be investigated.It will be shownthat streamwisevorticity generated by the Stokeslayercanenhancethe receptivityof the meanflowboundarylayerto free-stream disturbances. The downstreamevolution of the vortical motion of the fluid will also be modeled and the results will be presented.[Work supportedby AnalogDevicesProfessorship.] The interaction between acoustic disturbances and the mean-flow THURSDAY MORNING, 19 NOVEMBER 1987 TUTTLE SOUTH ROOM, 9:00 TO 11:35 A.M. SessionKK. PhysicalAcousticsVI: GeneralTopicsin PhysicalAcoustics Michael E. Haran, Chairman IBM FederalSystems Division,Manassas,Virginia22110 Chairman's lntroductionB9:00 ContributedPapers 9:05 KKI. Meusurementof the universalgasconstantR usinga spherical acousticresonator.M. R. Moldover, J.P. M. Trusler, T. J. Edwards, J. B. Mehl, and R. S. Davis (ThermophysicsDivision, Center for ChemicalEngineering, NationalBureauof Standards, Gaithersburg, MD 2O899) A spherical acoustic resonator hasbeenusedto redetermine theuniversalgasconstantR with an uncertainty of 1.8 ppm (standarddeviation).To accomplish this,threesubtasks werecompleted. ( 1) Thevolumeofa spherical shellwasdetermined byweighing themercuryrequired to exactlyfill it at thetemperature of thetriplepointof water,273.16K. (2) With theresonator filledwithcommercially suppliedargon,theresonancefrequencies of the radialmodesweremeasured asa functionof pressure. Usingourtheoretical modelforthecavity,thefrequency measurementswere combined with the mean resonatorradius determined in subtask ( 1) toobtainthespeedofsoundin commercially supplied argon. (3) Finally,thespeedof soundin thecommerically supplied argonwas compared to thespeed of soundin a "standard" sample whosechemical andisotopic composition wasaccurately established. discovered. Previously, solidscouldbeclassified ascrystalline, withperiodiclatticespacing, orasglassy, withrandomsitespacing. Thenewquasicrystalstructures appearto havelong-range order,showing sharppeaks in theFouriertransformspaceasin a periodicsystem,but theyalsohave properties thatareimpossible foranyperiodic structure, suchasfivefold rotationalsymmetry. For periodicsystems, Bloch'stheoremmaybeused to understand physical properties suchaswavetransmission, andit isof currentinteresttolearnif anysuchsymmetrytheorems applytoquasiperiodicsystems. Rigorous theorems forone-dimensional quasiperiodic patternsbasedon a Fibonacci sequence suggest thatsucha patternmaybe usefulin control of vibration transmissionin rib-stiffenedplates.How- ever,in twoandhigherdimensions, littleisknownabouttheconsequences of quasiperiodic structure. Recently,acoustic measurements havebeen madeona two-dimensional quasiperiodic systemandthefrequency spectrum,densityof states,andeigenmode patterns,showing somefeatures uniqueto thequasiperiodic pattern,havebeendetermined. [Worksupported,in part,by NSF DMR 8701682andthe Officeof Naval Research.] 9:50 KK4. Interaction of a soundbeamwith a two-fluidinterface.Jacqueline 9:20 KK2. Effects of carbon dioxide and humidity on some acoustic and thermodynamicpropertiesof air. George S. K. Wong (Division of Physics,National ResearchCouncilof Canada,Ottawa, Ontario K1A 0R6, Canada) The theoreticaldata on the combinedeffectsof humidity and carbon dioxidecontentonsomephysicalproperties of air, suchasthecharacteris- ticimpedance pcandthesound speed c,thespecific heatsCpandC,, and their ratio y, and the densityp, havebeenstudied.In general,overthe temperature range0•-30øC,thenormalized values C•,/(C•, and C/Co,becomelargerwith the inclusionof humidityand risingtemperature;and they are alsoinverselyproportionalto the CO2 content. Similarly,P/Po andpc/( pC)oare proportionalto the CO2 content,but they are inverselyproportionalto humidity and temperature.However, thenormalizedvaloe7//7/obecomes smallerwith theincrease of humidity, temperature,and CO2 content.The abovereferencevalues,whichare indicated witha zerosuffix, suchas(Cr) oandpo, refertodrystandard air Naze Tj•tta (Departmentof Mathematics, The Universityof Bergen, 5007 Bergen,Norway, and Applied ResearchLaboratories,The Universityof Texasat Austin,Austin,TX 78713-8029),Hanne Sagen (Departmentof Mathematics, The Universityof Bergen,5007Bergen, Norway), and SigveTj•tta (Department of Mathematics,The Universityof Bergen,5007 Bergen,Norway, and Applied Research Laboratories,The Universityof Texas at Austin, Austin, TX 787138029 ) The reflection and transmission of a real sound beam at the interface betweentwo homogeneous and dissipative fluid layersare considered. Numericalresultsare obtainedby usinga fastFouriertransformalgorithm.For thetransmitted field,theyshowthat,at a givenincidentangle, the directionanddisplacement of thebeamdependcriticallyon the absorptioncoefficient, andonthedistance betweenthesourceandtheinterface.Variousasymptotic formulasarealsopresented, whichallowfor a physicalinterpretation of thenumericalresults. 10:05 (314 ppmCO2content)at 0 øCandat a pressure of 101.325kPa. KKS. The role of cutoff modes in waveguides with boundary 9:35 KK3. Eigenmodesof quasicrystals.J. D. Maynard and Shanjin He (Departmentof Physics, The Pennsylvania StateUniversity,University Park, PA 16802) Recently,a newstateof matter,referredto as quasicrystalline, was S70 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 discontinuities.R. Sen (Department of Engineering Science and Mechanics,Virginia PolytechnicInstitute and State University, Blacksburg,VA 24060) and CharlesThompson(Departmentof ElectricalEngineering, TheUniversityof Lowell,Lowell,MA 01854) Whena waveguide witha boundary discontinuity isexcitedby a lowfrequency planewave,cutoffcrossmodesareinduced by thediscontin- 114thMeeting: Acoustical Society ofAmerica $78 uity. Althoughthesemodesare discernible only in the vicinityof the discontinuity, theyact as sourcesin the incompressible localflow,and therebyaffectthemeanacoustic pressure. Thishasa profoundeffecton the farfieldreflectionand transmission coefficients R and T, and, conse- quently, onthejunctionimpedance Z. Themethod of matched asymptoticexpansions shallbeusedandit will beshownthat,byproperlyaccounting for the crossmodes,the frequency rangeof validityof asymptotic estimates of R, T, andZ canbe greatlyextended.The possibility of improvingtheseestimates bytheuseofPad6fractionswill alsobediscussed. Finally, the formal difficultiesof incorporatingcutoff modesinto the acousticfielddescription arediscussed, andan extension of themethodof matchedasymptoticexpansions to circumventthem is presented. 10:20 KK6. Hot-film anemometerresponseto hydroacousticparticle motion under imposedbias flow. Pieter S. Dubbelday (Underwater Sound ReferenceDetachment,Naval ResearchLaboratory,P.O. Box 568337, Engineering and Ferroelectrics Laboratory,INSA, 69621Villeurbanne Cddex,France) Standardson piezoelectricity[IEEE Trans.SortiesUltrason.SU-31 (March 1984) ] describe theanalysis of vibrations on piezoelectric materials havingsimplegeometricalshapes.The resultsare basedon linear piezoelectricity, and resonance modesare treatedas noncoupled vibrationalmodes.However,in general,realmaterialsinvolvecoupledmodes. In addition,they may haveonedimensionthat is not greaterthan the others.Therefore,it is not possibleto stateconciselya specificset of conditionsunder which the definitionsand equationscontainedin the standards apply.However,for pure-mode directionpropagation, themechanicaldisplacements are only relatedto a singlecoordinate,and the electricalimpedance andgeneralelectromechanical couplingfactorequationscan be determined.With the sameassumptions as in the thickness excitationof plates[ M. Brissaud,J. Acoust.SOc.Am. SuppLI 79, S46 (1986) ], that is, completelystress-free bounda• conditionsandlossless piezoelectricmateddais, the impedanceequationof thediskcanbegiven.It shouldbenotedthat,whenthepiezoelectric coefficients c• andh3•vanish,the impedancereducesto an equationthat issimilarto that givenby Mason'sone-dimensional model [Phys. Acoust.I ( 1964}]. Orlando, FL 32856-8337} The response of a hot-filmanemometer to verticalhydroacoustic particlemotionisinfluencedby freeconvection,whichactsasa biasflow.The outputwasshownto be proportionalto particledisplacement for a wide rangeofparameters [ P.S. Dubbelday, J.Acoust.Soc.Am. 79, 2060-2066 (1986)]. It wasexpectedthat an imposedbiasflowwouldincreasethe outputandremovethedependence onthedirectionofgravity.Therefore, a hot-filmsensor(diameterd) wassubjected to an underwaterjet from a nozzle.The ratio of the voltageoutputs,due to a hydrostaticfield,with and without this dc flow, wasmeasuredas a functionof frequencyfor variousflowspeeds. The outputshoweda transitionfrom beingproportional to particlespeed,to beingproportionalto particledisplacement, dependingon the angularfrequencycoand imposedflow speedv. The transitiontakes place when a dimensionless number f•, definedas 1• = r•l/v, isoforder1.Thisphenomenon hasconsequences for theinterpretationof the measurement of turbulence,whereit iscustomaryto use the dc velocitycalibrationwithout referenceto frequencydependence. [Work supported by theONR. ] 11.•5 KK9. Photoacousticeffectfrom fine particles:Grain size and frequency dependence of the acousticyield. J. Pelzl,U. Netzelmann(Insfitut fdr Experimentalphysik, Ruhr Universit•it,D-4630 Bochum 1, Federal Republic of Germany), and D. Schmalbein(Bruker Analyfische Messtechnik GmbH, D-7512 Rheinstctten, Federal Republic of Germany ) The audioacoustic responseof air in contactwith finecopper-sulfate particlesIhat are heatedperiodicallyby •sonant microwaveabsorption hasbeeninvestigated asa functionof the modulationfrequency( 10Hz to 5 kHz) of themicrowavepowerfor differentparticlesizes( 10p to I mm ). Absolutevaluesofthephotoacoustic (PA) yieldarcobtainedbynormalizationwith the simultaneously detectedconventionalEPR signal.The PA-signalgeneration efficiency hasbeenfoundto increase stronglywith decreasing particlesizewith the maximumvalueoccurringat a distinct size-dependent frequency. Thesizeandfrequency behaviors of theamplitudeand phaseshiftof the PA-signalare analyzedtheoretically on the basisof a modelthat considersa singleisolatedparticle. 10.'35 KK?. On the extensionof the acousticv•riationalprinciple,from media at restto potentialmeanflows.L. M. B.C. Campos(InstitutoSuperior T•cnico,1096LisbonCodex,Portugal) The well-knownacousticvaddationa] pddnciplc for soundin a medium at rest [e.g.,Morseand Ingard,Theoretical ,4coustics (McGraw-Hill, NewYork, 1968},p. 248],is generalized to theacoustics of a potential meanflow,andisusedto derive:(i} theenergyequationfor soundin f•c and ductedflows;(ii) the waveequationin an inhomogeneous fluid at rest,in low Mach numberandin high-speed convection. {iii) the Web- 11:20 KKI0. Extrapolation of acousticfields by meansof source function estimation. G. Elias and F. Payen (Department of Physics,Office National d'Etudes et de RecherchesA6rospatiales,BP 72, 92322 sters'hornwaveequations andwaveequations for low Maeh numberand Chatilion Cedex, France) high-speed nozzles.Case Off) aboveappliesto quasi-one-dimensional soundpropagationin nonuniformducts,and generalizes the variational Extrapolationof acousticpressures is neededwhen farfieldmeasureprincipleofWeibel [J. Acoust.SOc.Am. 75, 105-106( 1955}] to include - mentsare not possible.Most of the methodsusuallyusedare basedon a the effectsof potentialmeanflow. The linearacousticwaveequations direct applicationof Kirehhoff'sintegralor on spectraldecomposition, derivedfromthevariafional principlearechecked by comparison with whichleadsto a finesamplingof a closedsurface.The presentmethod exactnonlinearequationsobtainedby a distinctmethod. requiresonlya limitednumberof microphones andconsists of a source functionestimation.In the firststep,the microphonearray is usedas a focusedantennafor measuringthe overallextentof the sourceregion.In thesecondstep,a biorthogonal basisiscomputedfor theexpansion of any sourcefunctionlimited to the previousspatialextent and the pressure radiatedon the array. Then, by projectingthe measuredpressureon this 10:50 KKS. Three-dimensiooal analysis for radial and thickness mode excitation of piezoceramic disk. Michel Bddssaud (Electrical S79 d. Acoust. Soc. Am. Suppl. 1, Vol. 82, Fall 1987 basis, anestimate ofthesource function canbeobtained toextrapolate the acousticpressurein the farfield. Simulationsshow a good agreement betweenreal and estimateddirectivitydiagramsand the preciselimitationsof this method. [ Work supportedby DCN. ] 114th Meeting: Acoustical Society of America S79 THURSDAY MORNING, 19 NOVEMBER 1987 UM AUDITORIUM LOBBY AND PREFUNCTION AREA, 9:00 A.M. TO 12:00NOON Session LL. Speech Communication V: Perception (PosterSession) Linda Polka, Chairman Department of Communicology, CB•4229, University of SouthFlorida,Tampa,Florida33612 ContributedPapers All posters willbedisplayed from9:00a.m.to 12.00noon.Toallowcontributors theopportunity toseeother posters, contributors ofodd-numbered papers will beat theirposters from9:00to 10:30a.m.,andcontributors of even-numbered paperswill beat theirposters from 10:30a.m.to 12:00noon. LLI. Individualdifferences in theperception of cuesfor initialstopplace and voicingcontrasts.Valerie Hazan (Departmentof Phoneticsand Linguistics, UniversityCollegeLondon,4, Stephenson Way, London NW1 9DY, United Kingdom) Fifteennormal-hearing adultsweretestedon theirabilityto identify synthesized stimulifromthreeinitialstopcontrasts. Twoplacecontrasts, BAIT-DATE andDATE--GATE, werepresented in threeconditions.In the "full-cue"condition,the placecontrastwassignaled by changes in burstfrequencyand F2/F3 transition.In the two "reduced-cue"condi- ticms, thecontrasts weresignaled byonlyoneof thesetwocues.A voicing contrast,GATE-KATE, waspresented in two conditions: a 'Tull-coe" condition,in whichthe contrastwasmarkedby changesin VOT andFI onset;anda "VOT cue"condition in whichtheFI onsetwasfiat.Group resultsshowsignificantvarianceacrossconditions,both in termsof the phonemeboundaries and identification functiongradientsobtained.In addition,individualdifferences werefoundin thelabelingof reduced-cue conditions. Somesubjects showed littledifference between thelabelingof the full-cue and the reduced-cue conditions of the BAIT-DATE and CATE-KATE contrasts, whileothersweretotallyunableto establish a contrast for some conditions. syllablesare usedasstimulusmaterials.Nonsenseconsonantclusterscon- stitutea classof sounds that arelargelydevoidof semantic context,yet theyrepresent a higherlevelofphonological organization. Consequently, therecognition of thesesoundsmayreflectprocessing levelsthatarenot involvedin theperception of simplerstimuli.In thepresentstudy,consonant clusters(CCV or CCCV in/a/vowel context), as well as their singletone members, werepresented to listenersagainsta background noiseat 10, 5, and 0 dB S/N'. The resultsshowsystematictrendsin terms of the frequencyand the patternsof recognition errors.In general,the proportion ofsubstitufions, additions, anddeletions resemble thepattern of errorsobservedin productionerrorsof consonant clusters.Theseand furtherresultswill bepresented anddiscussed at themeeting. LIA. Auditoryfactorsin theperception of stopsandglides.Margaret A. WalshandRandyL. Diehi (Department of Psychology, University of Texas, Austin, TX 78712) Thereissomedisagreement in theliteratureaboutthe relativecontri- butionofformanttransition durationandamplitude risetimeincuingthe stop--glide distinction (e.g.,[hi vs [w] ). Lackof resolution on thispoint ispartly dueto the confoundingofthesetwo variablesin moststudies.For a seriesof identification experiments, setsof synthetic[ba] and [wa] LL2. Perception of S-to-unvoieed-stop transitions in natural and synthetic speech. JanP.H. vanSantenandBeverlyM. Glasgow(AT&T Bell Laboratories, Murray Hill, NS 07974) The perceptualroleof the preclosure portionof S-K, S-T, andS-K .transitionswas investigated.Mahalanobisdistancesbetweenreflection coefficientvectors(derived from linear predictivecodinganalysis) showedthatdifferences between thestopsarcaslargebeforetheclosureas during the release.In a paired-comparison experimentusingnatural speech,listenerspreferredconsistent words (e.g., S-T followedby TOUT) overinconsistent words(e.g.,S-K followedby TOUT). However,perceptual differences weresubtle,andthestopidentitywasneverin question.In a subjective qualityratingexperimentusingsyntheticspeech, there were no differences betweenconsistentwordsand stop-neutral words(S-silencefollowedby, e.g.,TOUT), but inconsistent wordswere givenlowerratings.It isconcluded thattheperceptual roleof thepreclosureportionofS-to-anvoiced stoptransitions issmall,andcanbeignored in currentspeech synthesizers. Thisapparent conflictbetween theexperimentalresultsand the Mahalanobisdistanceanalysismay involvethe perceptualsalleheyof the dimensions alongwhichvectorsdifferand the effectsof amplitude.Finally, the resultsdemonstratethe sensitivityof subjective qualityratings. LL•. Ree•nition of consouantclusters.Moshe Yuchtman (Project Phoenixof Madison,Inc., 2001S.StoughtonRoad,Madison,W153716), ]oseph Stemberger (Department of Linguistics, University of Minnesota, Minneapolis, MN 55455),andChristopherMartin (Speech ResearchLaboratory,Departmentof Psychology, IndianaUniversity, Bloomington,IN 47405) Perceptualprocessing of speechhasbeenfrequentlystudiedusing analyses of consonant recognition errors.Typically,nonsense CV or VC S80 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 stimuli were created in which transition duration and rise time varied orthogcmally. Bothvariables affected labelingperformance in theexpecteddirection, buttransition durationwasbyfarthemoreimportantfactor. A similarpatternof resultswasobtainedfor singlesine-wave stimulithat modeledthe risetimes,frequencytrajectories,and durationsof the first formantin the [ba] and [wa] stimuli.A likelyauditorybasisfor boththe speech andnonspeech effectsisthatrapidfrequency transitions andshort risetimesbothcontributeto a higherdegreeof spectralsplatter(i.e., dispersion of energyacross a widefrequency band),whichmaybea principalcuefor abruptonsets in general.[Work supported by NICHD.] LLS. Identificationof stopconsonants from continuousspeechin limited context. Lori P. Lamel (Room 36-545, Department of Electrical Engineeringand Computer Science,and ResearchLaboratory of Electronics,Massachusetts Institute of Technology,Cambridge,MA 02139) Thispaperdescribes a series ofperceptual experiments aimedat evaluatinglisteners'abilityto identifysingletonstopconsonants. The stimuli consistedof portionsof speechextractedfrom a corpusof about 3600 sentencesspokenby over 4.50talkers.A variety of factors,suchas the stresspattern,syllableposition,andphonemicenvironment,weresystematically varied to determinethe extentto which thesefactorsinfluencethe listeners'decision.At least20 listenersheardsyllable-initialstopsin one taskandnonsyllable-initial stopsin another.In eachtask,onlythe immediatesurroundingcontextwasavailable.Overall identificationrateswere betterthan97% forsyllable-initial stopsand85% forthenoninitialstops. In thesyllable-initial case,75% of theerrorswerein voicing,predominantlyvoicedstopsbeingheardas their voiceless counterparts. For the noninitialstops,88% of theerrorswerealsoin voicing.However,in this task,mostof theerrorswerevoiceless stopsheardasvoiced.Almost70% oftheerrorsforthenoninitialstopsinvolved alveolars, whereas nopartic- 114thMeeting: Acoustical Society ofAmerica S80 ular placeof articulationaccountedfor mostof the errorsfor the initial stops.Factorsotherthansyllablepositionwill alsobediscussed, andcomparisons will bemadetootherstopidentification studies.[ Worksupported by DARPA undercontractN00014-82-K-0727,monitoredthrough the Officeof Naval Research.] LL6. The roleof rateof transitionin placeperception. ReneeA. E. Zakia and John Kingston(DMLL, Morrill Hall, CornellUniversity,Ithaca, NY 148534701 ) Kewley-Port,Pisoni,and Studdert-Kennedy["Perceptionof static anddynamicacousticcuesto placeof articulationin initialstopconsonants," J. Acoast. Soc. Am. 73, 1779-1793 (1983)] showedthat initial bilabialand alveolarstopscanbe reliablyidentifiedfrom informationin the first20 msof the syllable,includingthe burstand the beginningof the transition.Identificationof stopsas velarsrequireda longerstimulus, indicatingthat a slowchangein formantfrequenciesis characteristicof this placeof articulation.If rate of changeof formantfrequencies is a positivecueto theidentification of velarstops,it wouldbepredictedthat varyingrate of changewouldaffecttheir identification.To testthis prediction,syntheticstop-vowel stimuli,derivedfrom naturaltokens,consistingof all combinations of [b,d,g] andthe vowels[i,a], wereeditedto producestimuliwithfourratesof formanttransition:100,75, 50,and25 ms.Listenersin thisexperimentidentifiedplaceofarticolationconsistently regardless of rate of transition;specifically,there wasno significant interactionbetweenplaceandrateof transition.This wouldindicatethat a slowformanttransitionis not necessary ira stopis to be identifiedasa velar.There was,however,a significantinteractionbetweenvoweland placeof articulation,whichimpliesthat themagnitude,if notthe rate,of the transilionis a significantcueto place. LL7. Rate versus duration of formant transition as a discriminative cue. Robert $. Porter, Jr. (Kresge Hearing Research Laboratory, Department ofOtorhinolaryngology, LSUMedicalCenter,NewOrleans, LA 70112and Departmentof Psychology,Universityof New Orleans, New Orleans, LA 70122), Jane Collins, John K. Cullen, Jr. (Kresge Hearing ResearchLaboratory,Departmentof Ortorhinolaryngology, LSU Medical Center, New Orleans, LA 70112 and Department of Communication Disorders,Collegeof Arts andSciences, LouisianaState Universitj', Baton Rouge, LA 70804), and Dena F. Jackson(Kresge Hearing ResearchLaboratory, Department of Otorhinolaryngology, LSU Medical Center, New Orleans, LA 70112) The linguisticmessages of speech cannotbeclearlyreadin itstime-by- frequency-by-amplitude signature norisit manifest in thecorresponding psychoacoustic domain.It issuspected that importantaspects of speech messages may, instead,be revealedin higher-order,rate-of-change attributesofspeech signals andtheircorresponding perceptual representation. In a recentexplorationof rate-of-change-of-spectrum [Besinget al., J. Acoust.Soc.Am. Suppl. 1 gl, S54 (1987) ], 300-ms,second-formant-like signalswete used,whichwerevariedin onsetfrequencyto producevaria- tionsin therateof a 30-ms,initialtransition.Resultssuggested listeners may userate-of-frequency-change to discriminateamongthesesignals. The presentstudyexaminedthresholdsfor rate changesproducedby transitiondurationvariationinsteadof onsetfrequency. Transitionsbeganat either 1500or 2100 Hz and roseor fell to a steadystateof 1800Hz. Standard transition durations of 60 and 124 ms were used. Discrimination wascomparedfor risingandfallingtransitions presented eitheralone,or presented togetherwith first-formantlike resonances. Resultsfor tensubjectssuggest, asin theprevious study,thatlisteners mayuseperceived rate changerather than perceiveddurationto discriminatethesespeechlike signals.[Work supportedby NIH-NINCDS and LouisianaLionsFoun- resultshavenot beenobtainedwhen natural adaptorswere employed [Summerfieldet al., J. Phonet. 8, 491-499 (1980); Shuster and Fox, ASHA 28, 72 (1986) ]. In the presentstudy,subjectsproducedexamples ofthetoken[rat"i] priortoadaptation andthenafterperceptual adaptationwitha naturallyproduced [raphi].Voiceonsettimewasthenmeasuredforeachspoken utterance. !t wasfoundthatVOTsweresignificantly shorterafter adaptationto the bisyllablethan they were prior to adaptation.Theseresultsare similarto thoseobtainedwith synthetic adaptorsandsupportthenotionthatperceptuomotor adaptationreflects a perception-production link. LL9. Central and peripheral representationof whisperedand voiced speech.Arthur G. Samuel(Departmentof Psychology, Box 11A Yale Station, New Haven, CT 06520) Whisperedspeechis verydifferentacoustically fromnormallyvoiced speechyet listenershavelittle troubleperceivingwhisperedspeech.Two selective adaptationexperiments exploredthebasisfor thecommonperceptionof whisperedand voicedspeech,usingtwo/ba/-/wa/continua (onevoiced,onewhispered).The firstexperiment usedtheendpoints of eachseriesas adaptors,and severalnonspeech adaptorsas well. Speech adaptorsproduced reliablelabelingshiftsof syllables matchingin periodicity (i.e.,whispered-whispered or voiced-voiced); somewhat smallerelfeetswerefoundwithmismatched periodicity. A periodicnonspeech tone withshortrisetimeproduced adaptation effects likethosefor/ha/. These shiftsoccurredfor whisperedtestsyllablesas well asvoiced,indicatinga commonabstractlevelof representation for voicedandwhisperedstimuli. Experiment2 replicatedandextendedexperimentI, usingsame-ear and cross-earadaptationconditions.There wasperfectcross-eartransferof the nonspeech adaptation effect,againimplicatingan abstractlevelof representation. Theresults supporttheexistence of twolevelsof processing for complexacousticsignals.The cornmortality of whispered and voicedspeech arisesat thesecond, abstract level.Boththislevel,andthe earlier,moredirectlyacoustic level,aresusceptible to adaptationeffects. [Work supported by AFOSR.] LLI0. Modeling the perception of vowel-like formant peaks. Hector Jaykin, Brian Hanson, Paul Neyrinck, and HisashiWakita (SpeechTechnologyLaboratory,3888StateStreet,SantaBarbara,CA 93105) An improvedmodeloftherelationship ofharmonicstructure to vowel perception is proposed andevaluated. Jaykin,Hermansky, andWakita Illth InternationalCongress of PhoneticSciences, Tallinn, Estonia, USSR(1987)] compared listeners' responses toone-formant stimuliwith the computedweightedaverageof the two mostprominentharmonics [ themostimportantfrequency, or MIF of Carlson,Font,andGanstrom, •4uditory,4nalysis and Perception ofSpeech(Academic,London,1975)] appliedwithdifferentscaling factors.With thismeasure, a relativelyexpandedscalesuchasmagnitudecomesclosest to perceptual testresults. To improveon the characteristics of MIF, a criticalbandanalysiswas adoptedand Chistovichand Chernova [SpeechCommon. 5, 3-16 (1986)] werefollowedin applyinga centerofgravityformantestimation (CO). Because CG is highlydependent on the integration interval,an iterativeprocesswasused,with eachiterationchoosingthe frequency limitsfor the nextanalysisuntil convergence. Initial resultssuggest that, whileCO is lessaffectedby amplitudeexpansions thanM1F, themagni- tudespacemostcloselyapproximates listeners'responses for the oneformant stimuli. Evaluations of the method with multiformant stimuli are alsopresented. dation. ] LLII. Acousticversusphonologiealcontexteffectsin the perceptionof formant transition continua. Terrarice M. Nearey and Sherrie LLS. Perceptuomotor adaptation to natural speech adaptors. Linda 1. Shuster (Department of Speech Pathology and Audiology, West Virginia University,Morgantown,WV 26506-6122) E. Shammass(Departmentof Linguistics,Universityof Alberta, Perceptuomotoradaptationhas beendemonstratedusingsynthetic adaptors[Cooperand Nager, J. Acoust.Soc.Am. 58, 256-265 ( 1975); JamiesonandCheesman,J. Phonet.15, 15-27 (1987) ]. However,similar articulationin formant transitioncontinuais perhapsthe archetypalex- S81 J. Acoust.Soc.Am. Suppl.1, VoL82, Fall 1987 Edmonton T6G IE7, Canada) The effectof the followingvowelon listener's judgmentsof placeof ampleof contextsensitivity in speech perception. Is thissensitivity explicablein termsof acousticcontextaloneor mustthejudgedphonological 114thMeeting:AcousticalSocietyof America S81 identityof thefollowingvowelalsobeconsidered? Shammass [unpub- ated.Next, a sensory-perceptual transformation is appliedto generate a lisheddissertation,Universityof Alberta (1985) ] designedtransition- unitaryperceptual path.Sincethesensory-perceptual transformation is reactive, theperceptual pathovershoots theburst-friction onsetof the tokenandentersa physically unrealizable octantof APS,whereinthe perceptual targetzones forthestopsarelocated. Distincttargetzones are continuaexperiments involvingambiguous and nonambiguous vowel stimulito examine thisquestion in detail.Linearlogisticanalysis of these dataconfirms thesystematic roleofacoustic contexteffects, butindicates thatthephonological contexteffects areofa highlyrestricted natureand, furthermore,are both task and listenerdependent.A summaryof this analysis is presented andtheresultsarediscussed in thecontextof other resultsfrom the literat. ure (includingthoseof Mermelstein,Massaro, Ohala,andRepp). The evidence is consistent with the hypothesis that phoneme-sized units(ratherthandiphones or largerunits)serveasthe primaryinterface between acoustic cuesandphonological units. estimated for thestops[p,t,k,d,b,g],for h-likeaspiration, andfor voice bars.[Supported by AFOSR andNINCDS.] LLI$. Locationsof the burst onsetsof stopsin the auditory-perceptual space. Allard Jongman (CentralInstitutefortheDeaf,SaintLouis,MO 63110) LLI2. Identification of "hybrid" vowels in sentencecontext. James J. Jenkins and Winifred Strange (Department of Psychology, Universityof SouthFlorida,Tampa,FL 33620) It hasbeendemonstrated that listenerscanidentifytheintendedvowel in a CVC syllableevenwhenthe vowelnucleushasbeenattenuatedto silence,leavingonlyongiides andoffglides(silentcentersyllables).VerbruggeandRakerd [ Lang.Speech29, 39-57 (1986)] constructed "hybrid" silentcentersyllablesby crosssplicingongiidesand offglidesof citation-formCVC syllablesspokenby a maleanda femaletalkersuch thatformanttrajectories werediscontinuous. Identification oftheintendedvowelin hybridsyllables wasnolessaccuratethanvowelidentification ofthesingle-talker silentcentersyllables. Thepresentstudyreplicates this researchwith syllablesspokenin a carrier sentence,"I say the word /dVd/some more," usingten American-English vowels.Hybrid silent centersyllables werepreparedby crossing thestimulussentences sothat PreliminaryLPC analyses of thevoiceless stopconsonants [ p,t,k] in thecontextof thevowels[i,•,t,a,u,u] producedin realCVC wordsby one maleandonefemalespeakersuggest a methodfor locatingformantsin burst-friction spectra.Specifically, a 24-msHammingwindowwascenteredover the onsetof the burst.First, the spectralpeakwith maximum amplitude below6 kHz waslocated.Then,movingfromleftto right,the centerfrequencies of the firsttwo peakswithin10 dB of the maximum werepickedastheburst-friction components BF2andBF3.Furthermore, in thosecaseswhereBF2 hadbeenpickedandBF3 wasseparated from BF2by 2500Hz or more,thefrequency valuefor BF2wasalsousedfor BF3. Theseburst-friction prominences werethenconverted intocoordinates of the auditory-perceptual space:x=Iog(BF3/BF2) and z = 1og(BF2/SR),whereSR is the sensoryreference. This methodof analysis enabledusto distinguish thestops,in termsof placeof articulation,with goodaccuracy.[Work supported by AFOSR.] the sentence started with one talker and switched to the other after the silentportionof the testsyllable.Silentcenterstimuliwerepreparedfor eachtalkerseparately ascontrols.Vowelidentification wasasaccuratein thehybridconditionasin thecontrolconditions suggesting that dynamic syllablestructureprovidestalker-independent informationaboutthearticulatory/acoustic event.[Researchsupportedby NINCDS.] LLI3. Classification of vowelproductions by meansof perceptualtarget zones: A response toLaderoger andStuddert-Kennedy. James D. Miller (Central Institutefor the Deaf, SaintLouis,MO 63110) In discussion ofmypaper[Miller,J. Acoust.Soc.Am. Suppl.81,S16 (1987)] at the 113thMeeting,Dr. Ladefoged andDr. Studdert-Kennedy requested specific numbers regarding percentcorrectclassifications of vowelproductions achievable throughuseoftheauditory-perceptual theory andits perceptual targetzones(PTZs). At the timeof the 113th Meeting,theestimated shapes andlocations of thePTZswereverypreliminaryandrepresented onlya second iteration(I2). Thepercents correct for the I2 zoneswere moderate (50%-95%). Basedon reconsideration of older data and on numerous new measurementsmade in our laboratory, a newsetof PTZsis beingdeveloped for the vowels.These represent thethirditerationor I3 zones. TheI3 zoneswill bepresented andpercentages of correct classifications forseveral datasetswillbereported.Issues involved in developing decisive testsof theperceptual targetzones will bediscussed. [Supported byNINeDS andAFOSR.] LLI6. Signal detection analyses of speaker identification accuracy. {3eorgePapeun(Los A!amosNational Laboratory,Los Alamos,NM 87545),Jody Kreiman (PhoneticsLaboratory,UCLA, 405 Hilgard, Los Angeles,CA 90024), and Anthony Davis (Departmentof Linguistics, StanfordUniversity,Palo Alto, CA 94305) The accuracy of speakeridentification canbemeasured undervarious experimentalparadigmsincluding:(a) multiple-choice,closed-setexperiments,in whichlisteners chooseonespeakerfromamonga lineupofn speakers[as in F. Mc{3ehee,"The reliabilityof the identificationof the human voice," J. Gem Psychol. 17, 249-271 (1937)]; (b) multiplechoice,open-setexperiments, in whichlistenerschooseonespeakerfrom amonga lineupof n speakers,but can alsorefuseto chooseany of the speakers[as in C. P. Thompson,"Voice identification: Speakeridextiffability and a correctionof the record regardingsex effects,"Human Learn. 4, 19-27 (1985)]; and (c) independent-judgment, open-setexperiments,in whichsubjects are to makeeachchoiceindependently of others[asin {3. Papcun,J. Kreiman,andA. Davis,"Memoryfor unfamiliar voicesat delaysof one, two and four weeks,"submittedfor publication]. In this paper,it wasdemonstrated that signaldetectionanalysis, with certainextensions, makesit possible to compareresultsacrossthese diverseexperimentalparadigms,andto seethe essentialidentityof apparentlydisparateresults.A computerprogramis providedto treatcase(b). LL17. Long-term changesin voice characteristics:Implications for speakeridentity verification(SlV). Timothy C. Feustel and {3eorge LLI4. Auditory-perceptual approachto the stop consonants. James A. Velius (Bell D. Miller and Allard Jong•nan(Central Institute for the Deaf, Saint Morristown, Louis, MO 63110) Usingnaturaltokensandthe synthetic stimuliof Abramsonand Lisker[ 6th Internat.Congr.Phonet.(1970)], the auditory-perceptual theoryshallbeapplied tosyllable-initial stopconsonants. Theburst-fric- Communications Research, 435 South Street, NJ 07960) In a typicalspeakeridentityverification(SIV) scenario, a talkerfirst assertsan identityclaim via somemeansotherthan speech.A reference sampleof the claimedindividual'sspeechis thenloadedinto a pattern marcherandcompared againsta newsampleof thesamewordor phrase tioncomponents of eachtokenareanalyzed for thelocations of their solicitedon-line. Basedon the similarity of the two utterances,and a second andthirdsensory formants, BF2andBF3.Thesearelocatedin the predetermined criterion,a decisionis madeas to the veracityof the claimedidentity.One difficultproblemfor SIV applications is that the spectral characteristics ofa speaker's voicechange overtimein waysthat arepoorlyunderstood. Utterances werecollectedfromthreespeakers at regularintervalsovera periodof 23 weeks,andwereanalyzedfirstwith respect to themagnitude ofspectral dissimilarity overvaryingtimeintervals,and, second,with respectto the effectof thesedifferences on error auditory-perceptual space(APS) by theformulas: x = 1og(BF3/BF2) andy = 1og(BF2/SR),whereSR is thesensory reference. Theglottalsourcecomponents arethenanalyzed for theirsensory formants: SFI, SF2,andSF3.Thesearethenlocatedin APSby equations: x = 1og(SF3/ SF2),z = 1og(SF1/SR),andz = log(SF2/SF1 ). In thisway,a sensory pathcomprised ofburst-friction pointsandgiottal-souree pointsisgener- $82 J. Acoust.Soc:Am.Suppl.1, Vol.82, Fall1987 114thMeeting: Acoustical Societyof America $82 ratesfor an SIV decisiontask.The data showan increasein meanutter- antedissimilarity of about8% overa periodof 2 weeks, followed bya muchmoregradual increase over6 months. Whiletheabsolute change in meandissimilarity wassmall,it resulted in a doubling of theerrorratefor sharper temporal onsets andoffsets, andwithanenhanced dynamic range in thefrequency dimension. [Work supported by DARPA underContractN00039-85-C-0254, monitored throughNavalElectronics Systems Command.] the SIV decisiontask from about 5% to 10%. LL18.Auralsignalprocessing in thecontextof a digitalhearingmodel. D.M. Chabries, M. W. Christiansen, R. W. Christiansen, R. H. Brey, R. W. Harris (BrighamYoung University,Provo,UT 84604), and M. Robinette(Mayo Clinic,Rochester,MN 55905) A nonlinear model ofthehuman auralsystem isproposed. Utilizing a homomorphic transformation, themodelisimplemented ina digitalcomputerandaccounts fortheFletchercriticalbandsin hearing, thenonlinearrelationship between inputsoundintensity andperceived loudness knownasrecruitment, aswellasthethresholds ofaudibility andthemaximumtolerance threshold. Thetransformed or outputsignaldomainis identified asa perceptual space. It isshownthatthemodelcanbeusedto reproduce theFletcherMunsonequalloudness contours fornormalhearingpopulations. Modification ofeasilymeasured parameters results in the generation of equalloudness contours for hearingimpairedindividuals. Througha combination oftheforwardhomomorphic transformation for normalhearing totheperceptual domain andtheinverse mapping utiliz- ingparameters ofa hearing impaired individual, it isshown thathearing intelligibility isdramatically improved. It ispostulated thatsignal processing performed in theperceptual domainwill produce results more pleasing andintelligible tothelistener. Results ofauditory testing utilizingtheproposed modelforbothnormalandhearing impaired populationsarepresented. Speech intelligibility scores areshown toincrease by morethan40% (i.e.,from40% toover80%) overpreprocessed scores underheadphones forhearing impaired subjects usedin thesetests. LLI9. A modelforthetransduction stageofauditoryspeech processing. StephanieSeneft(Room 36-549,ResearchLaboratoryof Electronics, Massachusetts Instituteof Technology, Cambridge, MA 02139) A newmodelforthetransduction stage ofauditory processing isproposed. Themodelisappliedindependently toeachof40criticalbandfilter outputs,spanningthe frequencyrangefrom 100-6400Hz. The model consists offourcomponents in cascade: aninstantaneous nonlinear half- waverectifier, a short-term adaptation component, a low-pass filter,and, finally,a rapidadaptation component. Modeloutputswerecompared LL21, Alternatemeasures of response biasin speech recognition by older listeners.Robin T. K. Zcllcr (Speechand Communicative Disorders, Adelphi University,Garden City, NY 11530) and LouisJ. Gerstman (Departmentof Psychology, City College,CUNY, New York, NY 10031) Fifty-twolisteners rangingin agebetween 50and81years,listened to two R-SPIN Tests,responding to eachstimulusverballyandwith a numericalconfidence rating.Twelveperformance measures werecreated: for HP andLP itemsseparately, percentcorrect,d' andP(A ), whichboth measuresensitivity,and B, which measuresresponsebias;for all 100 items,a meanandstandarddeviationof theratings;andfor theerrorson theLP items,a meanfrequency of wordusage, anda ratingofcontextual feasibility bya panelof 15judges.Correlations withagerevealed noeffects fortheHP items,butestablished thatwithadvancing age,therewaslower sensitivity ontheLP itemsasmeasured byd' andp(A), butnochangein B. Additionally,the olderlistenersuseda narrowerrangeof the rating scaleandmademorecontextually feasible errors.No agevariationwas foundforwordfrequency oferrors,butall theproposed responses were tentimesmoreprobablethanthemisheardLP words. LL22. Perceptual and acoustic effects of nonlinear signal processing. Moshe Yuchtman,Rick Jenison,and Robert Lasky (Project Phoenix of Madison,Inc., 2001S. StoughtonRoad,Madison,WI 53716) A commonstrategyfor accommodating reduceddynamicrangein the hearingimpairedistheuseof amplitudecompression. In orderto address the resultingelevationof the noisefloor, an expansionregionmay be introduced.This study examinesthe perceptualand acousticconsequences of suchnonlinearsignalprocessing of CV andVC monosyllables. Transmittedinformationfor a systemof distinctivefeaturesasa function of thedegreeof nonlinearity, S/N ratio,andspeakerwasevaluated using theSINFA procedure. Acousticanalyses of processed syllables werecarried out in an attemptto definearticulationindexmeasuresthat account for the perceptualdata. with availableauditorydata in a numberof differentdimensions as fol- lows:( I ) onset response totonebursts, (2) forwardmasking effects, (3) steady-state behaviorfor vowel-like stimuli,(4) incremental response characteristics, and(5) synchrony lossat highfrequencies. Twodistinct spectral representations arecomputed fromthemodel outputs: anamplitudespeclrum corresponding t6 mean-rate response anda synchrony spectrum whichmakesuseoftemporal information to enhance spectral peaks. Themodeliscurrently beingincorporated intoa speaker-independentcontinuous speech recognition system underdevelopment. [Work supported by DARPA underContractN00039-85-C-0254,monitored throughNavalElectronics Systems Command. ] LL20. Pulse synchronous analysisof speechusing an auditory representation. JamesR. Glassand StephanieSeneft(Room 36-585, Departmentof ElectricalEngineeringand ComputerScience,and Research Laboratory of Electronics, MassachusettsInstitute of Technology,Cambridge,MA 02139) A procedure hasbeendeveloped that automatically extractspulse excitation information fromthespeech signal.Thistime-domain algorithmoperates on thetransduction stageoutputsof an auditorymodel developed bySeneft [thismeeting]. Specifically, robust inflection points aredetermined froma waveform obtained bysumming thechannel outputs.Thi• pulselike representation provides information aboutperiodic andaperiodic excitation, andcanbeusedtoextractperiodicities in voiced regions. Thustheinformation provided by thissignalcomplements the spectral information provided bymean-rate response outputs oftheauditorymodel.Furthermore, thissignalcanbeusedto samplechanneloutputspulsesynchronously. This produces a spectralrepresentation with 883 J. Acoust. Soc.Am.$uppl.1, Vol.82, Fall1087 LL23. Longitudinalchangesin syllablediscriminationamongnormaland handicappedschoolchildren. Lois L. Elliott and Michael A. Hammer (Audiology and Hearing Impairment, Northwestern University, Evanston, [L 60201 ) Cross-sectional age-related differences in discrimination of voiceonset time (VOT) [L. L. Elliott, J. Acoust.Soc.Am. $0, 1250-1255(1986)] and the place-of-articulation (POA) feature [L. L. Elliott et al., J. Acoust.Soc.Am. 70, 669-677 (1981)] whenjust-noticeable differences amongfive-formant synthesized CV syllables are measured usingadaptive testprocedures werepreviouslydescribed. Resultsare reportedhere for childrentestedlongitudinally for 3 consecutive years.Discrimination of temporalinformation(VOT) showedlargerchangesover time than discrimination of frequency-based information (POA). Also,largerperformancedifferences occurredbetweennormallyprogressing children andthosewith learningdisabilities for VOT thanfor POA. Childrenwith learningdisabilitiesgenerallyexhibitedpoorerauditorydiscrimination thannormals;however,theydemonstrated asmuchlongitudinal change in performances asdidthenormallyprogressing children(i.e.,therewere no significantinteractions betweenyearof testandnormalversushandicappedstatus).[Supported, in part,by NINCDS andNIH.] LL24. The role of static spectralpropertiesin vowel identification. James Hillenbrand and Brendan J. McMahon (RIT Research Corporationand RochesterInstitute of Technology,75 Highpower Road, Rochester, NY 14623) The classicstudyof Petersonand Barney [J. Acoust.Soc.Am. 24, 175-184 (1952) ] demonstratedthat vowelsin CVC context are identified 114thMeeting: Acoustical SocietyofAmerica $83 with a very high degreeof accuracyby humanlisteners.A numberof attemptshavebeenmadeto classifythePetersonandBarneyvowelsusing somecombination ofF0 and the threelowestformantfrequencies. Althoughtheseattemptshavebeenreasonably successful, the accuracyof theclassification algorithmshasbeenwellbelowthat of humanlisteners. Humanlisteners, however, haveaccess todynamic( durationandspectral change)aswellasstaticinformation(F0 and"target"formantfrequencies). The purposeof the presentstudywasto determinethe identifiability of the Petersonand Barneyvowelsetbasedexclusivelyon staticinformation. A formantsynthesizer wasusedto generateisolated,steady-state versions of all 1520vowels(76 speakers X tenvowelsX tworepetitions) inthePeterson andBarneydatabaseusingmeasured valuesofF0 andF IF 3.Preliminaryresultsfroma smallgroupoftrainedlisteners suggest that the error rate for the resynthesized stimuliis 3.5-4.5 timesgreaterthan the error rate reportedin the original Petersonand Barneystudy.Pilot dataalsosuggest thattheidentification taskmaybeextremelydifficultfor someuntrainedlisteners. [Work supported by RomeAir Development Centerundercontract• F3060285-C-0008.] (1987) ]. Acousticanalysis of thecategorized syllables revealednosingle featurethat couldhave supportedgeneralization, suggesting that the quailusedsomecombination of stimulusproperties to categorizesyllables.Currentexperiments investigate twowayssucha category couldbe formed.First,initialconsonant informationmaybeevaluatedconditional uponthevowelthatfollows.Whetherquailcanlearnto categorize syllablesin which, for example,positiveexemplarsare CVs with eithera voicedstopanda frontvowel,or a voiceless stopanda backvowelisbeing tested.A secondpossibilityis that phoneticcategories are examplesof polymorphous concepts, with no singlenecessary or sufficient condition for classmembership. Two birdshavelearneda conceptin whichthree stimulusdimensions of syllablesare neededto makea categoryassignment.The bestsubjecthaslearneda conceptthat requiresherto discriminate: (1) voicedversusvoiceless syllable-initialstops,(2) front versus backvo?els,and(3) sexofthetalker.Responses tonovelstimulireveala categorystructureremarkablysimilarto that hypothesized for humans. The fact that quailcanlearna conceptin whichmultiplecuesareclearly requiredsuggests that generalmechanisms of auditionand perceptual learningare sufficientfor acquisitionof speechcategories.[Work supportedby NICHD. ] LL25. Vowelandconsonant judgmentsare notindependent whencuedby the sameinformation.D. H. Whalen (HaskinsLaboratories,270 Crown Street, New Haven, CT 06511 ) Vowelswith ambiguous formantpatternscanbeperceived aschanging from [•e] to [5] as durationdecreases. Similarly,ambiguous final consonants canbeperceived as [d] whenthepreceding vowelislongand It] when it is short. Mermelstein[Percept.Psychophys. 23, 331-336 (1978) ] had subjectsidentifyfour syllables("bad," "bat," "bed," and "bet") thatweresynthesized byvaryingonlyvowelduration.He claimed theconsonant andvoweljudgmentswereindependent, implyingdistinct vowelandconsonant systems. The presentreplication indicates thatMermelstein'sanalysislackedpowerand that, in fact, thejudgmentsare dependenton eachother:Subjects weremorelikelyto reporthearing[•e] and [t ] together,sincemoreof thevocalicsegment durationisattributed to thevoweland/or lessisattributedto theconsonant. Similarly,subjects are morelikely to reporthearing[•] and [d] together,sincethereis less duration attributed to the vowel and/or more attributed to the consonant. Thisindicates thatlisteners heara coherent syllable,in whichtwodistinctionsthatdependonthesameinformationmustcompetefor thatinformation. [Researchsupported by NIH Grant HD-01994.] LL28. Qualitative separatenessin children's speech.Susan Nittrouer (Haskins Laboratories, 270 Crown Street, New Haven, CT 06511 and Boys Town National Institute, 555 North 30th Street, Omaha, NE 68131) and D. H. Whaien (Haskins Laboratories,270 Crown Street, New Haven, CT 06511 ) Acousticanalyses of fricative-vowel (FV) syllables spokenby adults andyoungchildren(ages3 to 7 years)havedemonstrated severaldifferences.In children'sspeech:( 1) thegrossspectraof/]'/and/s/segments weremoresimilarthanthoseof adults;( 2 ) fricativeF2 frequencies varied moreasa functionofvowelcontext;and(3) therelativeF2 amplitudewas greater.Theseacoustic differences betweenchildren'sandadults'speech providedan opportunityto testtwocompeting hypotheses of perceptual segmentation. Accordingtoone,listeners dividethesignalintotemporally discrete,context-sensitive allophonesthat can be usedto makeinferencesaboutlater-arrivingsegments. The second hypothesis suggests that listenerscan separatesimultaneously arrivingacousticinformationinto successive phoneticunits.In thepresentsetofexperiments, briefportions of fricativenoiseextractedfrom FV syllablesspokenby adultsand childrento normaladultlistenerswerepresented. Resultsshowedthat listen- LL26. Statisticalrepresentation of word-initialobstruents:Further data. Karen Forrest,(3ary Weismet,and Paul Milenkovic (SpeechMotor ControlLaboratories, Universityof Wisconsin Madison,Madison,WI 53705-2280) At the 113thMeetingof theAcousticalSociety,a procedurewaspresentedfor the classification of word-initialobstruents. The procedureinvolredcalculatingFFTs at ! O-msintervalsfrom theonsetof theobstruent throughthe third cycleof the followingvowel.TheseFFTs weretreated as random probabilitydistributionsfrom which the first four moments (mean,variance,skewness, andkurtosis ) werecomputed; computations were also made on Bark transformed FFTs. Moments calculated over the first40 msof voiceless stopsdiscriminated placeof articulationwith97% accuracy.Voiceless sibilantswerecategorizedwith about98% accuracy when the Bark transformed moments from the first 20 ms of the fricatives were used.The speechsampleusedin the previousstudyonly allowed comparison ofobstruents across twovowelcontexts. In thisreport,results of the momentsanalysisof a speechsample are presentedin which all voicedandvoiceless stops,aswellasvoiceless sibilants,arepairedwith six vowels,/i,a,u,•e,n,e/. The efficacyof the momentsin discriminatingobstruentsin this broadercontextwill be discussed. [Work supportedby NIH award NS13274.] erswerelessaccuratein fricativeidentification for child speakers, but moreaccuratein identifyingthe vowelcontextfrom whichthe fricative noisecame.Thusthegreatercoproduction foundin children's speech was moreeasilyrecovered by listeners, eventhoughthemostprominentphone was more poorly perceived.This findingwas taken as supportfor the secondhypothesis.[Work supportedby NIH GrantsNS-07237and HD01994 to Haskins Laboratories.] LL29. Recognitionof speechand recognitionof speakersex:Parallel or concurrentprocesses? M. O'Kane (School of Information Sciencesand Engineering,Canberra College of Advanced Education, Belconnen, A.C.T. 2616, Australia) If the samephoneticstringis spokenby a malespeakerand a female speaker,bothdrawn from a linguisticallyhomogeneous population,then a listenerperceivesthat both speakersare sayingthe samesoundsbut generallyknowsthat the speakersare differentand, in particular, that the speakersare of a differentsex.What is not well understoadis whetherthe listenerhasto determinethe identityof the sexof the speakerin order to determinetheidentityof thesounds,or whetherthe two piecesof informationare essentially independent andcanbe determinedconcurrently, Japanesequail.Keith R. Kluenderand Randy L. Diehi (Departmentof Psychology, Universityof Texas,Austin,TX 78712) or whetherthe identityof the soundscanbe roughlydeterminedwithout knowledgeof the speaker'ssexbut finerecognitionis possibleonly after the speaker's sexis known.In thispresentation, evidencefor andagainst thesevariouspossibilities froma largenumberofpsyeholinguistic studies, Japanese quail (Coturnixcoturnix)havelearnedto form a category for syllable-initiai[d], and candiscriminate[d] syllablesfromthosebeginningwith [b] or [g] acrossvowelcontexts[Kluenderet aL, Science particularlythosedealingwith varioussex-specific effectsin a varietyof languages, is surveyedand the implicationsfor the designof automatic speechrecognitionand understanding systems are discussed. LL27, Use of multiple speech dimensions in concept formation by S84 J. Acoust.Soc. Am.Suppl.1, Vol. 82, Fall 1987 114thMeeting:AcousticalSocietyof America 584 LI30, Effectsof speakergenderon speechrecognition by hearing impaired listeners. AngelikaStieren and WinifredStrange (Department of Communieology, University of SouthFlorida,Tampa, LL31. Are female voicesmore intelligible than male voices?Florion J. Koopmans-van Beinum and Mirjam T. J. Tielen (Institute of PhoneticSciences, Universityof Amsterdam,Hcrcngracht338, 1016CG FL 3362(I) Amsterdam, The Netherlands) Audiologists andclients oftenclaimthatfemale speech islessintelligible than malespeech for subjects with bilateralhigh-frequency sensorineuralhearinglosses. To examinetheextentto whichsourcecharacteristics(voicepitch)andgender-related resonance characteristics contribute to thesespeechrecognitiondifficulties,two men and two womeneach produced threesublists of themodified rhymetest(MRT). Withingendergroups, onespeaker hada relativelylowvoicepitch,onea highpitch, suchthatthehigh-pitched malevoicewasapproximately thesameasthe low-pitched femalevoice.Stimuliwerepresented to listeners at 30 dB SL in quiet,in speechnoise(S/N = q- 12 riB), and in speechbabble(S/ to studythe relativeintelligibilityof maleand femalevoices,ten maleand ten femalespeakers wereaskedto readthreephonotactically balanced listsof 50 CVC combinations. Specialcare wastaken with respectto speech level,signal-to-noise ratio,andpresentation methodusing:equiva- N = q- 12dB). Overallresultsshowedsignificant differences in identifi- cationofconsonants across speakers. However,contraryto expectations, performance wasworstforthemalewiththelowvoicepitch,intermediate for the two femalespeakers, and bestfor the high-pitched male.There appeared to begender-related differences in thetypeof consonant errors. THURSDAY MORNING, 19 NOVEMBER Studies on theintclligibility of femalecompared to malevoices, and studieson the relationbetweenacousticparameters (suchasFo) and speech understanding areratherscarce andindirect.However,particular]y in information-providing services, femalevoicesare frequentlyused, whereas,in speech-technology applications,they arc often excluded, probablydueto inadequate methods foranalyzingfemalevoices.In order lent peaklevel,babblenoise,and a variationof the Nakatani and Dukes method.For a smallnumberof trainedlisteners, theintelligibilitythresh- oldperspeaker isestablished, providing20 measuring pointsperlistener. Furthermore,wordandphoneincintelligibilityperspeakeraremeasured with phoneticallybalancedsentences andwords,at a fixedS/N ratio. 1987 TUTTLE NORTH ROOM, 9:00 A.M. TO 12:00 NOON SessionMM. StructuralAcousticsandVibrationIV: ViscoelasticPolymers:Their Descriptionand Evaluation Wayne T. Reader, Chairman Code1905.2,David W. TaylorNavalShipR&D Center,Bethesda, Maryland20084 Chairman's Introduction--9:00 Invited Papers 9:08 MM1. Viscoelasticityof polymers•Consideratlonof predictivemethodsfor wide frequencyresponse. RaymondE. Wetton(PolymerLaboratories Limited,The Technology Centre,EpinalWay, Loughborough LEI 1 0QE, United Kingdom) Thenatureof viscoelasticity in highpolymersisdiscussed in molecularterms.The factorsaffecting dampinganddynamicmodulus areoutlined.Manycommercial polymers aremultiphase andthewidthandmagnitudeof molecularrelaxationareshownto be relatedto the compatibilityof the components. Mastercurves showing viscoelastic response overa widerangeoffrequency canbegenerated usingsemiempirical procedures of time-temperature shiftingof data.The Williams,Landel,and Ferry (WLF) relation,in particular,is discussed for multi-andsingle-component polymers. In general,predictions of dynamicresponse viaextrapolationoveroneor twodecades in frequency canbemadewithreasonable confidence, if enoughdetailedcareis taken.The useof dielectriclosspeakloci at highfrequencies is proposedto aid thesepredictions. 9:30 MM2. Thetemperature dependence ofviscoelastie responses above Ts. RobertF. Landel(JetPropulsion Laboratory, 4800 Oak Grove Drive, Pasadena,CA 911091 Theoriginsof theWLF equation will bereviewed, alongwiththeroleof thetemperature dependence of responses in tryingtoascertain relaxation mechanisms. Somepitfalls,misuses, andmisunderstandings will be described, together withsimilarcomments onso-called Ts's.Thegeneral applicability andutilityofthetemperature scaling ofthetimedomainwillbeillustrated withexamples fromdamping, rupturebehavior, friction, and crack propagation. $85 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America $85 9:55 MM3. Molecularbusisof •he acousticpropertiesof imlyurethanes. Bruce Hartmann, Gilbert F. Lee, and JamesV. Duffy (PolymerPhysicsGroup,Naval SurfaceWeaponsCenter,SilverSpring,MD 20903-5000) Theacoustic properties of polyurethanes asa functionof frequency aredominatedby a verybroadrelaxationprocess in whichthepolymermakesa transition fromtherubberytotheglassy state.It isconvenient touse an analyticrepresentation of thedatathatexpresses thecomplexshearmodulus in theform G* = G• - (G• - Go)/[l + (itoh') •] , whereGoisthelimitinglow-frequency modulus,G, isthelimitinghigh-frequency modulus,r istherelaxation time,anda is a dimensionless parameter,with valuesbetweenzeroand one,that determinesthe width of the relaxation. It willbeshownthatr isrelatedto theglass transition temperature T• of thesoftsegment in the polyurethane. In turn, T• is foundto depend on thedegree of phaseseparation thatoccursbetween the chemicallyincompatible hardandsoftsegments, aswellasthedegreeof crystallinitydeveloped. Thesesame molecular factorsdetermineGo,butG, isfoundto beessentially constant. Attemptsto relatea to thewidthof theglasstransitionasmeasured in a differentialscanning calorimeterhavebeenonlyqualitativelysuccessful. [Worksupported by theNSWC Independent Research ProgramandtheOfficeof NavalResearch. ] 10:20 MM4. Techniquesfor noiseand signaturereductionapl•lications. D. J. Townend(Admiralty Research Establishment, Holton Heath, Poole,DorsetBH 16 6JU, United Kingdom) In recentyears,DMA testsystems havebecomecommercially available.Priorto this,workersin thisfielcl developed equipment to meettheirownspecific needs.Currentcommercial equipment isaimedat thethermal analysismarketandisseenascomplimentary to established techniques, suchasDSC andTMA. Greatemphasisis placedon temperature control,but,in general,lessattentionispaidto frequencyrangeandmechanical design.Upperfrequencylimitsare usuallyaround100Hz. While low frequencies are idealfor the studyof molecularrelaxationprocesses and for routinematerialcharacterization, they havelittle immediateapplication in the fieldof noiseand signaturereduction.Recentwork at ARE hasresultedin the development of computerized WLF technique;thismethodis basedon a referencefrequencyaswell asa referencetemperature.In general(thoughnotessential),thereference pointselectedisthetemperatureof peaktan at a frequency of I Hz. Wheremultiplexedfrequencydata are available,it is a simplematterto verifythat an unknown materialwill fit the standardequationby comparingmeasuredshiftsin peaktan with computedvaluesbased on the standardconstants.Signaturereductionapplicationsrequiresystemsthat, in somecases,work at frequencies upto I MHz. Computerized predictions basedonWLF data,shiftedfromlow-frequency measured data,showverygoodagreementwith high-frequency measuredacousticdata,but it wouldbe reassuring to be ableto measuredynamicmechanicalpropertiesat kilohertzfrequencies. Recentimprovements to an inertial mass,air-bearingsupported system,developed at ARE some10yearsago,haveresultedin a systemcapableof yieldingcomplexshearmodulusdata at frequencies up to l0 kHz. 10:45 MMS. Experimentalcoml•qson of dynamicmechanicaltestingfacilities at severalNavy laboratories.R. N. Capps,R. Y. Ting, and M. E. Quinn (Underwater SoundReferenceDetachment,Naval Research Laboratory,P.O. Box 568337,Orlando, FL 32856-8337) Dynamic mechanicalspectroscopy providesvaluableinformationon frequency-dependent relaxation mechanisms in highpolymers.The Young'smodulusandlosstangentare oftenmeasured,dueto the relative easeof experimental implementation. A varietyof techniques is used,whichoftengivesdifferentresultsfor samples that aresupposedly of thesamematerials.A problemexistsin interpretingsuchresultsin thatonecan neverbesureif suchdifferences are dueto measurement techniques or to actualmaterialpropertyvariations. Two systems,an automated resonancetechniqueand the Polymer LaboratoriesDMTA, are now used at severalNavy laboratories. To checkthe internalconsistency of resultsobtainedwith thesesystems, a "roundrobin"evaluationwasperformed.Commonbatchesof materialsof severaldifferenttypesof rubberswereused to preparesamplesfor eachof threedifferentNavy laboratories. The resultsof thistestwill be presented, and factorsthat canaffecttheaccuracyandreproducibility of dynamicmechanicalpropertymeasurements will be discussed. Criteriafor theapplicabilityof themethodof reducedvariableswill alsobetreated.[Work supported by NAVSEA. ] 11:10 MM6. Standardizedpresentation of complexmodulus.Lynn Rogers(AI•VAL/FIBA• Wright-Patterson, Dayton, OH 45433-6553)and Bryce Fowler (CSA Engineering,Inc., 560 San Antonio Road--Suite 101, Palo Alto, CA 943064682) The capabilities of a proposed standardfor graphicalpresentation of complexmodulusdatafor viscoelastic vibration dampingmaterialswill be described.A plot of material lossfactor versusmodulusmagnitudeindicatesdata quality;a frequency-temperature homogramis includedon this plot, therebyalsomakingit suffi- cient for the designer.The more typical reducedfrequencyplot is also presented,but with experimental temperatureandfrequencyrangesgivento assistthe designerin avoidingextrapolationof data.The stateof the art in modelingtemperatureshiftfunctionsandcomplexmodulususingfractionalderivativeswill bepresented.An interactivecomputerprogramfor processing a setof complexmodulusdata will bedescribed.A system to storeand retrieve dampingmaterial data will be described. S86 J. Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 114thMeeting: Acoustical Society ofAmerica 886 11:35 MM7. A computer-based systemfor representation of dampingpropertiesof materials.B. L. Burkewitz (BBN Labs,Inc.,Cambridge, MA 02138),J. L. Hamer(E-A-R Division,CabotCorp.,Indianapolis, IN 46268),andC. I. Holmer(NoiseControlTechnology, Inc.,Reston,VA 22091) Thedesignofa computer-based dampingdataanddesignsystem(DDDS) isdescribed. Importantfeatures of thesystemare:modularity,standardized interfaces betweenmodulesbasedon definedinformationcontent, easeofexpansion, physically basedrepresentation of materialdynamicmodulus,supportforall usersfromtest throughanalysis,andqualitycontrolto treatmentdesign.Systemarchitecture andmodulefunctionaredescribed,followedby a detaileddiscussion of the materialrepresentation module.The input for thismoduleis materialtestdata,andtheoutputisa setofanalyticexpressions forthefrequency andtemperature dependence of the material.The softwarefor thismoduleis basedon thework of Rogers[J. Rheol.27, 351-372 (1983) ], whichprovidesa rheologically consistent analyticfunctionrepresentation for the dynamicmodulusandloss factor.The moduleis writtenwithin RS/l, a table-oriented, scientificsoftwaresystem.Dampingtreatment designsoftwareand approaches to systemdataqualityissuesare alsoaddressed. THURSDAY MORNING, 19 NOVEMBER 1987 BRICKELL NORTH ROOM, 9:30 A.M. Meeting of Accredited StandardsCommittee S12 on Noise to be heldjointly with the Technical Advisory Group for ISO/TC 43/SC 1 Noise W. Melnick, Chairman S12 OhioState University,UniversityHospitalClinic,456 ClinicDrive, Columbus,Ohio43210 H. E. yon Gierke, Chairman,TechnicalAdvisoryGroup for ISOfFC 437SC 1 Director,Biodynamics & Bioengineering Division,AFAMRL/BB U.S.Air Force,Wright-Patterson Air ForceBase,Dayton, Ohio 45433 Standards Committee $12 onNoise.Workinggroupchairswill reportontheirprogress undertheplanfor the productionof noisestandards.The interactionwith ISO/TC 43/SC I activitieswill be discussed. S87 J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 114th Meeting:AcousticalSocietyof America •87 THURSDAY AFTERNOON, 19 NOVEMBER 1987 TUTTLE CENTER ROOM, 1:00 TO 3:50 P.M. SessionNN. PhysicalAcousticsVII: Acousticand SeismicCouplingII: Liquid-Solid Interfaces Michael A. Schoenberg, Chairman Schlumberger-Doll Research,OM QuarryRoad,Ridgefield,Connecticut 06877 Chairman's Introduction--l:00 Invited Papers 1:05 NNI. Seismic anisotropy in theupperoeeanle erust--Areview.R. A. Stephen (WoodsHoleOceanographic Institution, Woods Hole, MA 02543) Overthepast10years,therehavebeenfourreported observations ofseismic azimuthal anisotropy in the upper2 km of oceaniccrust.The mostprobable causeof anisotropy in the uppercrustis the preferred orientation of large-scale fractures thatarecreated duringcrustalgeneration at theoceanridges.Seismic velocities are slowparallelto the spreading directionandfastperpendicular to it. In someinstances, clear azimuthallydependenttravel-timeanomaliesare observedand, in others,thereis evidencefor shearwave splitting. At onesite,bothshear wavesplitting andtravel-time anomalies gaveconsistent results. Thechallenge in all studies, however, isto separate theeffects of anisotropy fromtheeffects of scattering andlateralhetero- geneity. Currentmodeling capability isinadequate toinclude theeffects ofbothanisotropy andheterogeneity atthescales observed intheseafloor. A definitive studyontheeffects ofanisotropy andheterogeneity thatwill determine whichobservations arebestforresolving givenparameters isneeded. In addition, high-quality threecomponent seismic stations arerequired to provide thedataforparticlemotionanalysis. Thismayrequire boreholeseismicinstallations.[Work supportedby NSF.] 1:35 NN2. Characterization of anisotropic elasticwavebehaviorin mediawith parallelfracturesandaligned cracks.MichaelSchoenberg (Schlumberger-Doll Research, OldQuarryRoad,Ridgefield, CT 06877-4108) Largeparallelfractures or aligned microcracks, whetherfluidfilled,empty,or debrisfilled,cause specific typesofmaterialcompliance increases, whicharemanifest in theanisotropic behavior of elasticwavesin such media. Theanisotropy attributable tothefracture system maybetriclinic(themostgeneral ) withsixindependentcompliances, monoclinic withfour,orthorhombic withthree,or transversely isotropic(rotationally symmetric fractures) withtwo.Thefracturesystem maybeembedded in ananisotropic elasticbackground withno restrictions on thetypeof anisotropy. To compute thelongwavelength-equivalent moduliof the fractured mediumrequires, at most,theinversion of two3X 3 matrices. Dilutesystems of flataligned microcracks, assumed onaverage rotationally symmetric, inanisotropic background yieldanequivalent medium of thesameformasthatof theisotropic mediumwithlargetransversely isotropic fractures; i.e.,therearetwo additional parameters duetothepresence ofthemicrocracks thatplayrolesin thestress-strain relations ofthe equivalent mediumidenticalto thoseplayedby theparameters dueto thepresence of largefractures. Thus knowledgeof the total of four parameters describing the restrictedtransverse isotropyof sucha fractured mediumtellsnothingofthesizeorconcentration ofthealignedfractures, butdoescontaininformation astothe overallexcesscompliancedue to the fracturesor cracksand their orientation. 2:05 NN3. Borehole modes in a fluid-saturated permeable medium: Numerical andexperimental results. Hsuilin LiuandKenneth W. Winkler(Schlumberger-Doll Research, Ridgefield, CT 06877) StoneIcy waves propagating inboreholes inpermeable materials provide a unique opportunity tostudythe boundary conditions ata fluid/permeable-solid interface. First,a review ofthetheory andpresent frequency domain solutions forStoneley wavevelocity andattenuation ispresented. Formation premeability isshown to decrease theStoneley wavevelocityandincrease theattenuation compared to thenonpermeable case.This effect ismostpronounced atlowfrequencies. Thenthetheory isapplied tofinitesizecylinders, suchasmight occurin laboratory experiments. In thiscase, twoclasses ofextensional modes, besides theStoneIcy mode,are found.Oneclassisthefastwavemode,whichisanalogous to theordinaryextensional mode.The otherclassis theslowwavemode, so-called because thezeros arelocated alongtheBiotslowwavebranch cut.Fora highly permeable medium, theslowwave modes canobscure theidentification oftheStoneley mode. Alsopresented is theexperimental confirmation of thenumerical predictions for Stoneley wavepropagation. Compared to conventional ultrasonic techniques, thenewStoneIcy wavetechnique allows theperformance measurements in a broadfrequency bandabove andbelowthecriticalfrequency ofBiottheory.Oneapplication oftheresults is estimating rockpermeability using theborehole Stoneley modein petroleum prospecting. S88 J. Acoust.Soc.Am.Supph1, Vol. 82, Fall 1987 114thMeeting:AcousticalSocietyof America $88 2:35 NN4. The effects of azimuthal anisotropyon seismicshear wave data. R. M. Alford (Amoco Production Research,P.O. Box 3385, Tulsa, OK 74102) As a resultof seismicfield experiments, shearpolarizationsplittinghas beenidentifiedas a significant impedimentto shearwave data quality. The polarizationsplitting,or shearbirefringence,resultswhen a transversely polarizedelasticshearwavepropagates throughan azimuthallyanisotropic medium.As a consequenceof thisobservation, the perception of shearwavepropagation hasbeensignificantly alteredandnew techniques for acquisition, processing, andinterpretation hasbeendeveloped. In thispaper,the anomalous behaviorof the dataisexaminedandtechniques aredeveloped to compensate for theeffectsof shearpolarizationsplitting.The correcteddatareveala time-varyingdelaybetweentwoshearpolarizations, whichischaracteristicof shearwavepropagation in an azimuthallyanisotropicmedium.Azimuthalanisotropymayprovide anotherparameterfor discriminating rock typesandproperties,and the possibilityof detectingfracturesand permeability trends.Therealizationthatshearwavesplittingisanimportantphenomenon helpsto explainthe erraticandunpredictable natureof sheardataquality.New techniques providean excitingrenewalof potential for theexploitation of shearwaveseismic datain exploration. ContributedPapers 3:05 NNS. Coupling of sound with surface waves on fluid-loaded elastic objects describedby a generalizationof GTD. Philip L. Marston (Departmentof Physics,WashingtonState University,Pullman, WA 99164) The geometricaltheoryof diffraction(GTD) wasrecentlyextended to describesurfaceelasticwave (SEW) contributionsof scatteringfrom fluid-loaded spheres andcylindersat highfrequencies [P. L. Marston,J. Acoust.Soc.Am. Suppl.1gl, S14 (1987) ]. The couplingoœan SEW with theacou:•tic fieldwasdescribed by a complexcoefficient G. That analysis gavesimplephysicalapproximations: IGl• 8,r•C/Cswfor spheresand ian beamrepresentation is introduced to replacethe traditionalplanewaverepresentation sothatthesolution isananalytical oneratherthana symbolic integral.Thisreplacement ismadepossible by a computer optimizationroutinethat overcomes problemsintroducedby useof nonor- thogonalfunctions, andenables theexpression of thefieldsolutionin a simpleclosed analytical form--thesumofGaussian functions. Thetechniqueisappliedtothedescription ofthereflected fieldofa bounded beam incidentat theRayleighanglewith verysatisfactory results.[Research sponsored bytheScience Alliance,a StateofTennessee Centerof Excellence.] ]GI• 8•'B/(•rka)•/2forcylinders; these wereconfirmed withothernumericalresults.The SEW angular attenuationcoefficient(due to radiation damping)and phasevelocityare/? and Csw,while c denotesthe soundspeedin thefluid,k = •o/c,anda istheobject'sradius.The present work extendsthe originalanalysisofbackscatteredsoundto caseswhere theSEW onlypartiallycircumnavigates theobject.If a planesoundwave ofamplitudepiexcites anSEWona cylinder,thesoundamplituderadiat- ed by the SEW aftertravelingan arc lengthL is p•p•(a/2r) •/• X 1G/21ex p( - •gL/a). Theobserver isa distance r froma virtualsource (in the cylinder), from whichthe radiatedraysappearto diverge.A simpleapproximation of/?showsthat/1octheratioof fluidto soliddensities. Theratioof the]G I forspheres andcylinders isconfirmed bygeometrical considerations. [ Work supportedby ONR. ] 3:20 NN6. A Gaussianbeam representationof wave fields at a liquid--solid interface.J. J. Wen and M. A. Breazeale(Department of Physics,The Universi. ty of Tennessee, Knoxville,TN 37996-1200) A newmethodof analysisisdescribed to representthe resultantfield whena wavebeamissubjected to anyphysicalprocess(e.g.,propagation, focusing,reflection,diffraction,etc.). Insteadof expressing thefieldsolution asa continuousdistributionof planewavesthroughspatialFourier analysis,a setof Gaussian beamsisemployedasthebasicunit.The Gauss- S89 J. Acoust.$oc. Am. Suppl.1, Vol. 82, Fall 1987 3:35 NN7. Saltus-typeboundaryconditionsfor the modelingof rough interfacesin acousticwaveproblems.AdriannusT. de Hoop, Sijtze M. de Vries (Facultyof ElectricalEngineering, Delft Universityof Technology, P.O. Box5031,2600GA Delft,The Netherlands), and DouglasMiller (Schlumberger-Doll Research,Old Quarry Road, Ridgefield,CT 06877-4108) The methodof movingvolumeaverages is usedto deriveboundary conditions of thesaltusor jump typefor the modelingof acousticwave interaction across a roughinterface between twomedia.Forthispurpose, a smoothboundary layerisconstructed thatcontains theroughinterface in its interior.The mediumin thislayerisconsidered asa mixtureof the twomediapresent. On thescaleat whichtheacoustic wavemotionis analyzed, themixtureisassumed tohaveacoustically linearmacroscopic properties. The resulting saltus-type boundary conditions express the jumpsin thebasicquantities thatdescribe thewavemotion(pressure or stress andparticlevelocity) across theboundary layerin termsof the limitingvalues thatthese quantities attainat eithersideof thelayerin a linearmanner.Conditions of thistypeareworkedoutfor a roughinterfacebetween(a) twofluids,(b) twosolids,and (c) a fluidanda solid.In allcases, fullanisotropy, including theonethatisinduced bytheinterface geometry, is takenintoaccount. 114th Meeting:AcousticalSocietyof America S89 THURSDAY AFTERNOON, 19 NOVEMBER 1987 UM AUDITORIUM, 1:00 TO 4:15 P.M. Session00. SpeechCommunicationVI: Focus Sessionon the Female Voice Caroline G. Henton, Chairman Linguistics Program,University of California--Davis,Davis,California95616 Invited Papers 1:00 OO1. The femalevoiceExperiments and overview.Gunnar Fant, ChristerGobl, Inger Karlsson,and QiguangLin (Department of Speech Communication andMusicAcoustics, RoyalInstituteofTechnology ( KTH ), Box 70014, S-100 44, Stockholm,Sweden) Resultsfromrecentexperiments at theKTH directedtothestudyof aerodynamics andvoicesourceflow will bepresented. Theseincludesimultaneous recordings of glottalairflowandsupraglottal pressure froma Rothenberg mask.A numberof females served assubjects producing isolated syllables and,in oneinstance, complete sentences. For onesubject, thesubglottal pressure wasalsorecorded. Typicaltemporalprofiles of these measures, including glottalleakage, willbediscussed. In a separate study,datawerecollected onglottal flowwaveforms frominverse filteringofthespeech pressure wave.A parametrization andresynthesis in terms of thefour-parameter LF modelhasbeenundertaken, whichallowsa comparison witha setof malevoices. Theoretical studies of source-vocal-tract interaction effects providea basisfor interpretation of resultsand modeling offemalespecific characteristics. Finally,a reviewispresented ofaccumulated insightin thepropertiesof femalespeechincludingfactorsotherthan sourcemechanisms. 1:30 002. Glottalairflowandpressure measurements forfemaleandmalespeakers in soft,normal,andloudvoice. E.B. Holmberg,R. E. Hillman (ResearchLaboratoryof Electronics,36-525, Massachusetts Instituteof Technology, Cambridge, MA 02139andDepartment of Communication Disorders, BostonUniversity, Boston,MA 02215), andJ. S. Perkell(ResearchLaboratoryof Electronics,36-525, Massachusetts Institute of Technology,Cambridge,MA 02139) Measurements ontheinverse filteredairflowwaveform (the"glottalwaveform") andofestimated average transglottal pressure andglottalairflowweremadeforproduction of syllable sequences in soft,normal,and loudvoiceby20femaleand25malespeakers. Statistical analyses showed that,whiletherewasnosignificant female-male difference in pressure withinanyof theloudness conditions, thereweresignificant female-male differences in theglottalwaveformparameters thatshouldberelevantto differences in voicesourcecharacter- istics.In normalandloudvoice,femalewaveforms indicatedlowervocalfoldclosingvelocity,lowerac (modulated) flow,andproportionally shorter closed phase ofthecycle,suggesting a steeper spectral slope for females. In softvoice,femaleacflowwassignificantly lowerthanmales',buttherewerenosignificant differences in female-male closing velocityor relativelengthof theclosedportion,suggesting moresimilarspectral slopes. Thedc (unmodulated) flowdidnotdiffersignificantly between females andmales. Possible implicationsof theresultsfor femaleandmalevoicequalityarediscussed. [Work supported by NIH.] 2.'OO 003. Physiology of the femalelarynx. Ingo R. Titze (VoiceAcousticsand Biomechanics Laboratory, Department ofSpeech Pathology andAudiology, University ofIowa,IowaCity,IA 52242andTheRecording andResearch Center,The DenverCenterfor thePerformingArts, 1245ChampaStreet,Denver,CO 80204) Anatomically,thefemalelarynxdiffersfromthemalein vocalfoldlength,thickness, angleof thethyroid laminae,restingangleof theglottis,verticalconvergence anglein theglottis,anda numberofotherways.The vocalfoldtissueitselfdifferswithrespect to thethickness of themueosaandfibercomposition, e.g.,theratioof connective tissue to muscle tissue. Biomechanically, it hasbeenfoundthatthestress--strain curves areslightly morelinearin females thanmales.Based onthese andotherknowndifferences in morphology andviscoelastic properties, a scalingprocedure isattempted for fundamental frequency, intensity,andotheracoustic properties.Somerationaleisgivealsoforwhyfemales develop pathology at differentsitesin thelarynx,givenwhatis currently known about mechanicalstressin the tissues. S90 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 114th Meeting: Acoustical Society ofAmerica S90 2:30-2:40 Break 2:40 004. Factandfictionin thedescription of femaleandmalepitch.CarolineG. Henton(LinguisticsProgram, Universityof California,Davis,Davis,CA 95616) Femalespeech hasbeencharacterized ashighpitched, shrill,overemotional, and.swoopy. Pitchis, of course,a relativescale,soit wouldbe equallyvalidto describe malepitchaslow pitched,rumbling,underexpressive, andmontonous. But if "swoopiness" is inherentin femalespeech, it mustbeaskedwhetherthe effectisoneof a comparatively widerrange,of greaterdynamicity, or of a simplearithmetical misinterpretation.It maybethecasethatprevious accounts havebeenbased onabsolute (linear)and,therefore, misleading valuesin Hertz;whereas theuseofthesemitonal, andthusperceptually moreappropriate, scale,indicates that female speech should notbesomarkedonacoustic bases alone.Reexamination ofpitchranges anddynamieity valuesreported byothersleadsto therejection of someof theelements of thestereotype. Newdataassembled fromfivemalesandfivefemales alsoindicatethat,auditorily,femalespeech contains similarratiosof pitch rangein relationto long-term average pitch,asdoesmalespeech. Monotonicity andreduced pitchrange,as important ingredients in enhancing stereotypical masculine speech (Terrangn,1974;BennettandWeinberg, 1974), are reexamined. 3:10 005. Acousticcorrelatesof breathiness:First harmonicamplitude,turbulencenoise,and trachealcoupling. DennisH. Klatt (Room 36-523, Massachusetts Instituteof Technology,Cambridge,MA 02139) A selected sampleof reiterantspeech hasbeencollected fromtenfemalespeakers andsixmalecontrols in orderto quantifyacoustic correlates of perceived breathiness of thefemalevoice,andto contrastthesemeasureswithcomparable datafrommales.Twosentences withdifferingstress patterns werespoken byreplacing eachsyllable by [hV] andby [•V], whereV = [a,i,ae,o,rr].Detailedanalysis ofthe [a] datareveals( 1) wide variationin thestrengthof thefirstharmonic(relativeto firstformantamplitude),withanaverage increase of about6 dB for femalesrelativeto males;(2) a greatertendencyfor the third formantto be excitedby noise ratherthanvoicingharmonics in thefemalepopulation; and(3) indications oftrachealpolesandzerosin the spectraof vowelsadjacentto voiceless consonants in utterances producedby bothgenders. Thesethreemeasuresofbreathiness tendto begreatest in unstressed syllables andtowardtheendofan utterance. The acoustic datawill becorrelated withperceived breathiness ofutterances fromindividualtalkersina preliminaryattempt to determinewhichmeasureis perceptually mostsalient;futureresearch usingsynthesis will systematically investigate perceptual sallenee. [Worksupported, in part,by an NIH grant.] 3:40-4:15 DISCUSSANTS: Sandra F. Disner R. A. W. Bladon THURSDAY AFTERNOON, 19 NOVEMBER 1987 UNIVERSITY LECTURE HALL, 1:15 TO 4:00 P.M. SessionPP. Psychologicaland PhysiologicalAcousticsVII: Masking and Detection Murray F. Spiegel,Chairman Bell Communications Research,435 SouthStreet,Room 2E-252, Morristown,New Jersey07960 ContributedPapers 1:15 PPI. Profile analysiswith time-varying stimuli. David M. Green and Quang T. Nguyen (PsychoacousticsLaboratory, Department of Psychology,Universityof Florida, Gainesville,FL 32611) Det•mtinga changein spectralshape,when the changeoccursat a singlespectrallocus,requiresa simultaneouscomparisonof the unchangeqpart of the spectrumwith the alteredor "signal"part. What are thetemporal parameters ofthestimulus thatipfiuence suchcomparisons? To considerthisquestion,a complexof 2l-equal-amplitude components $91 J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 wasused,rangingfrom 200-5000 Hz, with equallogarithmicspacing betweencomponents. The signalwasan increase in levelof onecomponentof themulticomponent spectrum.Overalllevelof thesoundvaried randomlyovera 20-dBrange.Thesignalcomponent andthe20nonsignal components wereamplitudemodulatedseparately. The relativephaseof the two modulatorswas manipulatedso that the signaland nonsignal components waxedandwanedeitherin-phaseor out-of-phase. For modulation rates faster than about 40 Hz, the thresholdsin the two conditions arenearlythesame.Thus,for slowerratesof modulation,thedatasuggest that thetemporalstructureof the spectrumis important,whileabove40 114th Meeting:AcousticalSocietyof America $91 Hz, onlythepowerspectrum of thestimulusis critical.[Research supportedby theNationalInstituteof HealthandtheAir ForceOfficeof ScientificResearch.] 1:30 PP2. Masker envelopesand simultaneous-masking psychophysical tuning curves. Todd W. Fortune and David A. Nelson (Hearing Research Laboratory,Universityof Minnesota,2630UniversityAvenue S.E., Minneapolis,MN 55414) Recentreportsdemonstrate thatthetemporalenvelope of a masking stimuluscaninfluenceitseffectiveness. To examinethe probableeffectsof maskcrenvelopeon psychophysical measures of tuning,simultaneousmaskingpsychophysical tuningcurves(PTCs) wereobtainedfrom two normal-hearing listeners.The probewasa 250-msl-kHz toneburstpresentedat 60 dB SPL, whichwastemporallycenteredwithin the 500-ms maskors.MaskorsincludedSAM and QFM tones,narrow-bandnoise, and puretones.PTCswereobtainedin quietandin the presence of a continuous broadband noise to mask combination tones. Results indicate that fluctuatingenvelopemaskersare lesseffectivethan flat envelope maskers onlyon thelow-frequency sideof the PTC. Thesedatasuggest thatthelistenerisableto listenfor thepresence of a probeduringperiods of low maskerenergy.Four of thesixmaskersinteractedwith theprobe producing audiblecombination tonesthatcouldbemasked bythebroadbandnoise.Combination tonesweremostpronounced for theQFM and pure tonemaskers, whichhavesimilarenvelopes. The significance of thesedatawill bediscussed in termsof frequency andtemporalresolution, particularlywithregardto thehearing-impaired population.[Work supportedby NINCDS. ] whichthe 295-mssignalintervalwaspreceded andfollowedby 295-ms presentations ofa sinusoid at thesignalfrequency. These"fringe"componentshadthesamephaseonall presentations; thesignalwasaddedto the maskerin quadrature.Signalthresholdwasmeasuredasa functionof the phaseanglebetweenthe fixedfringecomponents andthemasker.In non- signalintervals, themaskerandsignalwerereplaced bya presentation of thefringecomponent. Presentation ofthesignalthusproduced a change in both phaseand levelcomparedto the fringecomponents. However, thresholds werenotimproved bythepresence of thephase change. Both of theaboveconditions wererepeated fora 20-msmaskerandsignal,and againshowed noevidence thata change in phaseimproves detection performance. With thisshortersignal,it waspossible toexaminethreshold as a functionof the phaseanglebetweenthe signalanda 295-msforward masker.In forwardmasking,phasehadno effecton signalthreshold. [Researchsupportedby NSF.] 2:15 PPS.Maskingof shorttoneburstsof frequency sweeps. DavidA. Fabry, Margaret F. Cheesman(Department of CommunicationDisorders, Universityof Minnesota,Minneapolis, MN 55455),Neal F. Viemeister, and Dean Schei(Departmentof Psychology, Universityof Minnesota, Minneapolis,MN 55455) Maskedthresholds for a shortprobetoneare greaterfor sweep-frequencytonalmaskersthan for stationarysinusoidal maskersat equal SPL,andgreatest forsweep ratesof 20-30oct/s [G. F. Smoorenburg and F. Coninx,HearingRes.3, 301-316 (1980)]. The presentstudyexaminedpossible explanations for thissurprising phenomenon and concentratedon theeffectsof maskerlevelon probethreshold.Maskedthresholdswereobtainedfor a 10-ms,l-kHz tonetemporallycenteredin the 100-ms masker. Themaskers weresweptinfrequency atratesof0-80 oct/ 1:45 s and passedthrough I kHz at 50 ms. At the lower maskerlevels,maxi- PP3. Tuning curve widths of infants and adults in simultaneous and mummaskingoccurredat sweepratesof 15-25 oct/s. At higherlevels, little influence of sweeprateon maskedthresholds wasseen.Although nonsimultaneousmasking. Nancy BensonSpctncr and Lynne WernerOlsho(Departmentof Otolaryngology, R.L-30,Universityof Washington,Scattic,WA 98195) Tuning curve widthswcrc estimatedusingthe pulsationthreshold (PT) technique for infants,6-month-olds, andadults.The observer-based psychoacoustic procedure[L. W. Olshoet al., Der. Psychol.,in press (1987) ] wasthe testprocedure.Listenersdiscriminated a train of alternatingmaskerand probetoneburstsfrom a train of maskertonebursts presented with a continuous probetone.Thesestimuliwerepresented in a background of broadbandnoiseat 20-dBpressure spectrumlevel.The probelevelwas10dBSL.PTswereobtainedat thetuningcurvetip andat twopoints,oneaboveandonebelowtbeprobefrequency,10dB abovethe tip. The resultingQ•oswerecomparedto thoseobtainedin simultaneous maskingfor 6-month-oldsand adults[L. W. Olsho,Inf. Behar.Dev. 8, 371-384 (1985)]. Adult Q•osweresignificantly greaterfor PT tuning curvesat 500, 1000, and 4000 Hz. Infant PT •loS were significantly greaterthan simultaneous maskingQ•osat 500 and 1000Hz, but not at 4000 Hz. [Work supportedby NIH.] thereweresubstantial individualdifferences in slopes, the growth-ofmasking functions hadslopes lessthanonewithminimum slopes at20-25 oct/s.Theshallowslopes area distinguishing featureof sweep-frequency maskingandmayprovidea basisfor understanding thisimportantphenomenon.[Supportedby NINCDS grant NS12125and SSHRC-Canada.] 2:30 PP6. Tonal prominenceeffects on annoyanceratings for business equipmentspectra.G. R. Bienvenue(State Universityof New York, Collegeat New Paltz, New Paltz, NY 12561), M. A. Nobile, and A. C. Balant(IBM Acoustics Laboratory,Poughkeepsie, NY 12602) Dayton, OH 45435) Whilerecentresearchers havereportedAppendixB ofANSI S12.101985to "workwellin over80% of cases" andto be"a vastimprovement" overtheearliermethod,therearetwonotabledifficulties in itsimplementation.First,whentwo"audiblyprominent"tonesoccurcloselytogether withina singlecriticalband,theyarcratedas"notaudiblyprominent"by the presentAppendix,thoughthey producean audibleevent.Second, certainsharplyfallingnoisespectra could,bytheupwardspread ofmasking, maska tonethat is "audiblyprominent."In this case,an inaudible tonewouldberatedasprominent. Forthepresent study,15subjects were presented with tape recordedsoundsof varyingtone-noiseensembles designedto explorethesephenomena.Subjectsrated thesesoundsfor Thresholdsfor sinusoidalsignalsat 0.5, 2.0, and 6.0 kHz were measuredin a varietyof maskingconditionsusinga two-interval,forced- dataarediscussed withspecialattentionto theirrelevance forimplementingAppendixB of ANSI S12.10-1985.[ Worksupported by IBM Corpo- choiceprocedure.All conditionswereperformedin the presence of a diotic,white,broadbandnoisewith a spectrumlevelof 20 dB SPL. All sinusoidalstimuliwerepresented monaurallyin the listener'sleft car;all fixedlevelcomponents werepresented at 65 dB SPL. In a determination of theWeberfractionfor a 295-mssignalmaskedby a sinusoldof thesame frequencyand temporalcharacteristics, the thresholdvaried with the phasebetweenmaskerandsignal,asonewouldpredictfrom theassumptionthattbetaskdepends onlyuponthedetection of a difference in power (or relateddimension)betweenmaskcrand maskcrplussignal.In these data,thedetectionof phasedifferences, perse,appearedto haveno effect on threshold.Thesedata provideda referencefor a secondcondition,in ration.] 2:00 PP4. Phase has no effect on simultaneous or forward masked thresholds. DanielL. Weber(Departmentof Psychology, Wright StateUniversity, S92 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 their levelsof loudness,tonal prominence,and annoyance.The research 2:45 PP7. Modeling subjectresponses in a reproduciblenoisemaskingtask. R. H. Gilkey and T. A. Meyer (Signal Detection Laboratory,Central Institute for the Deaf, Saint Louis, MO 63110) Gilkey and Robinson[J. Acoust.Soc.Am. 79, 1499-1510 (1986)] modeledsubjectresponses to individualreproducible noisesamples, ran- 114thMeeting:Acoustical Societyof America S92 domly selectedon each trial from a set of 25, in a diotic tone-in-noise detectiontask. They found that a detectorcomposedof a 50-Hz-wide single-tuned filtercenteredat the 500-Hzsignalfrequency, a nonlinearity, andan integratorwith a 100-to 200-msdecayconstantpredictedthedata relativelywell. Here, a set of 150 noisesampleswas used,and a more detailedanalysisof the parameterspaceof the detectorwasperformed. The resultsof this studyare similar to their results,althoughthe bandwidthsol'the filtersare smaller(26-49 Hz) and the decayconstantsare shorter(39-100 ms). "Spectralweightingfunctions"werederivedfrom linearcombinations of the outputsof sevendetectors,eachat a different centerfrequency.Thesefunctionsare moreconsistent than thoseof Gilkey and Robinsonand alsoindicatethat subjectscompareinformation acrossfrequencyregions.Thesefunctionscan bedescribedasthe differencebetweentwo Gaussian-shaped weightingfunctions.[Supportedby NSF and AFOSR. ] the second patternare monotonically decreasing, whereasthosein the thirdpatternaremonotonically increasing. Sixexperimental conditions areexamined,obtainedbyvaryingthefrequency separation betweencomponentsandpatterns.The conditionwith thegreatestfrequencyseparation isonein whichthereis a one-fourthoctavegapbetweenpatterns;the conditionwith the leastfrequencyseparationis one in which all four patterns arecomposed froma singlesetoffrequencies. Exceptforthelast condition,successive patternswithineachconditionarecenteredaround 700, 1000,1400,and2000 Hz, respectively. Subjects'performance in the recognition taskis at the levelpredictedon the basisof their detection performance, in thoseconditions with greaterfrequency separation betweenpatterns. Recognition performance fallsbelowthepredicted level in thoseconditions withlittleor nofrequency separation. [Research supportedby AFOSR.] 3:00 3:30 PPS. Detection of infratonal repetition of frozen noise: Singularity recognition or linttern recognition? Brad fl. Brubaker and Richard PP10. Forced-choiceprocedurescan yield contaminatedthresholds: Musket resl•onsebias. Murray F. Spiegel (Bell Communications M. Warren (Department of Psychology,University of Wisconsin- Research,435 South Street, Morristown, NJ 07960) Milwaukee, Milwaukee, WI 53201 ) Listenerscanreadilydetectiterationof "frozen"Gaussiannoisesegmentswith durationsupto I s IN. GattmanandB. Julesz,J. Acoust.Soc. Am. 35, 610 (1963) ]. Is detectionof repetitionof thesecomplexwaveformsbaseduponrecognitionof the reoccurrence of uniquefeaturesor singularities, or upona moreholisticrecognition of thepatternformedby theseevents? To answerthisquestion,frozennoisesegments weredivided into threesectionsof equalduration(A, B, C) that werereassembled and thenrepeatedto form two periodicsounds(ABC), and (ACB),. This manipul•tionchangedthe temporalarrangementbetweensectionsbut preservedsingularitiesand repetitionrate. Untrainedlistenersheard a seriesof sequence burstsconsisting of eitheronearrangement [ (ABC),, (ABC),,.(ABC) ..... ] or two alternating arrangements[(ACB)•, (ABC)•,.(ACB) ..... ] and judged whether successive burstswere the sameor different.Discriminationwaspossiblewhenthe durationof the entireiteratedpattern(A 4- B •- C) was900msor less,indicatingthata holisticrecognition of patternsoperatesup to thelimit of echoicstorage. [Work supported by AFOSR.] Earlymeasurements ofhumanperformance wereplaguedbyresponse biases,suchasa preference for choosing"yes"in a yes/nosignal-detection task, or a tendencyto choosethe secondof two soundsas louder, longer,or higher.Signaldetectiontheory,as well as forced-choice and adaptiveprocedures, have largelyeliminatedthe influenceof response biaseswhile alsostreamliningthe data collectionprocess.Theseproceduresareadequatewhenthe setof maskersareindistinguishable. However,adaptiveprocedures, whencombinedwith forced-choice methods, are sensitiveto response biaswhenthe variousmaskersare distinguishable. The responsebias is a preferencefor particularmaskers,when maskersare randomizedwithin a trial. Thresholdscan be seriouslycon- taminated;depending onthemagnitudeof thebias,subjects canappearto be responding perfectlyto inaudiblesignals.Procedures that avoidthe problemresultingfrom thisformof response biasare not asefficientas adaptive,forced-choice procedures, and often entail more traditional methodsof testing. 3:45 3:15 PP9. Detectionandrecognitionof complexstimuli. David A. Ansleyand PPll. A context-free rubber band model of scalingauditory dimensions. Irwin Pollack (Mental Health Research Institute, University of DanielL. Weber(Departmentof Psychology, WrightStateUniversity, Michigan,Ann Arbor,MI 48109) Dayton, OH 45435) Greenet al. [J. Acoust.Soc.Am. 62, 948-954 (1977) ] havedemonstratedthat a subject'sability to detectoneof m possiblesignalsin a yes/ notaskcanbeusedto predictthesubject's abilityto identifywhichof the m possible signalswaspresentwhenthe signalsare four sinusolds well separatedin frequency.This studyexaminesdetectionand recognition whenthesignalsarefourcomplexstimulipresented againsta background of white noise(spectrumlevel20 dB SPL). Eachcomplexstimulusconsistsof a temporalsequence of sevensinusolds. Each sinusolddiffersin frequency fromtheotherswithina pattern,andeachispresented for 100 msat the samesensation level.The patternsdifferin the sequence of the frequencies of thecomponent sinusolds. For example,thefrequencies in $93 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 Auditorysignalsvaryingin frequency,duration,and/or amplitude werepresented tolisteners. Thelistener's taskwastomarka linearscaleto ratethe relativepositionof onesoundon a scaledemarkedby theother two.The signalsthatdefinedtheendsof thescalewerevariedfromtrialto-trial.Theratingsarewell-predicted byassuming thattheratioof scale distances is invariantwith the particularsoundsthat definethe scale. Statedotherwise,theratingscaleisconceived in termsof positions upona rubberband. Different scales,definedby differentend stimuli, resultin localcompression andextension of theband,butrelativedistances among intermediate ratingsareunchanged by scalechanges. The model'sapparentlyexcellentperformance may simplyresultfrom the powerfulconstraintof monotonicity in thescalingof unidimensional signals. 114thMeeting:Acoustical Societyof America S93 THURSDAY AFTERNOON, 19 NOVEMBER 1987 BRICKELL NORTH ROOM, 1:30P.M. Joint Meeting of AccreditedStandardsCommitteesS1 and S3 Theactivities ofSl willbediscussed first,proceeding tomatters ofinterest tobothSl andS3,andconcluding with S3 activities. Meeting of Accredited Standards Committee S1 on Acoustics D. Johnson, Chairman S l Larson-Davis Laboratories, 280 South Main, Pleasant Grove, Utah 84062 Standards Committee Sl onAcoustics. Workinggroupchairswillreportontheirprogress in thepreparation of standards, methodsof measurement andtesting,andterminology in physicalacoustics, electroacoustics, sonics, ultrasonics,and underwatersound.Work in progressincludesmeasurementof noisesources,noise dosimeters, integrating sound-level meters,andrevisionandextension of soundlevelmeterspecifications. Opendiscussion of committeereportsisencouraged. Meeting of Accredited StandardsCommittee S3 on Bioaeousties L. A. Wilber, Chairman S3 422 Skokie Boulevard, Wilmette, Illinois 60091 Standards Committee S3onBioaeoustics. Thecurrentstatus ofstandards underpreparation willbediscussed. In addition to thosetopics of interest including hearing conservation, noise, dosimcters, hearing aids,etc., consideration willbegiventonewstandards whichmightbeneeded overthenextfewyears. Opendiscussion of committeereportsis encouraged. Theinternational activities in ISO/TC 43 Acoustics andIEC/TC 29 Electroacoustics, for whichS1andS3 serveastheU.S.TechnicalAdvisoryGroups(TAGs), will bediscussed. TheChairsof theTAGs for ISO/TC 43 (H. E. vonGierke)andIEC/TC 29 (V. Nedzelnitsky) will reportoncurrentactivities of theseTechnical Committees. THURSDAY AFTERNOON, 19 NOVEMBER 1987 POINCIANA ROOM, 1:30 TO 2:45 P.M. Session QQ. Educationin Acoustics II: Educationvia SatelliteandVideoTapes Lawrence A. Crum, Chairman NationalCenter for Physical Acoustics, P.O.Box847, University, Mississippi 38677 Invited Papers 1:30 QQ1. A satelliteclassroomfor advancedacousticstudies.Richard Stern (Applied ResearchLaboratory, PennStateUniversity,P.O. Box 30, StateCollege,PA 16804) The GraduateProgramin Acoustics, PennStateUniversity,isofferinga Masterof Engineering Degreein Acoustics viasatellite totheNavyUndersea Weapons Engineering Station(NUWES), KeyportlWashington. Theprogramwill take2 yearstocomplete andconsists of bothtelevised andon-siteinstruction. Progress of the programand lessonslearnedwill be discussed. 1:55 QQ2. Grsduatecontinuing education by satellite.LionelV. Baldwin(NationalTechnological University, P.O. Box 700, Fort Collins, CO 80522) Progress in telecommunications technologies presents theengineering profession withexcitingnewwaysto integrate advanced studywiththejob.TenyearsagotheU.S.colleges ofengineering thatoperateregionalITV systems startedplanningfor a commongoal:to increase thenationaleffectiveness of continuing education of engineers. Two newinstitutions,the NationalTechnological University(NTU) and the Associationfor Media-BasedContinuingEngineering Education(AMCI•E) that coordinatecreditand continuingeducation programs,respective]y, are now accelerating the introductionoœlinkagesbetweenengineering facultyand S94 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting: Acoustical Societyof America S94 engineers atworkviaasatellite. BothNTU andAMCEEbegan regular satellite delivery ofadvanced technical education in August1985.Asweenterthethirdyearofoperations, NTU willservemorethan3200enrollees foracademic creditinstruction during1987-1988; AMCEE willprovide overa thousand hoursof noncredit, live-and-interactive programming. Theservice operates in GSTAR-Iwitha modern Kubandtransponder, whichbroadcasts twochannels offull-motion, colorvideothroughout thedayandevening ofeachworkday. Eighteen universities currentlyoperate uplinksto provideliveprogramming overthenetworks and,in addition,threecorporate studios haveuplinksandregularly originate instruction. 2:20 QQ3.Usingvideotapes andaudiotapesonandoffthecampus. Thomas D. Rossing (Department ofPhysics, Northern Illinois University,DeKalb, IL 60115) Theuseofprerecorded presentations offersseveral advantages to theteacher.Oneof themostattractiveis topresent demonstrations ofacoustical phenomena thataredifficult toimplement inrealtime,especially ina remoteclassroom. Anotheristosubstitute animated textforstilltext(theso-called "animated blackboard"). Excerptsarepresented fromseveralaudioandvideotapesandhowtheycanbeusedto enhance acoustics instructionbothon and offcampusis discussed. THURSDAY AFTERNOON, 19 NOVEMBER 1987 BOUGAINVILLEA ROOM, 1:30 TO 4:05 P.M. SessionRR. Engineering AcousticsIV: Transducers,Waveguides,and Structures TheodoreA. Henriquez,Chairman UnderwaterSoundReferenceDetachment,Naval ResearchLaboratory,P.O. Box 8337, Orlando,Florida 32856 Chairman's Introduction--l:30 ContributedPapers 2:05 1:35 RRI. Diistortionlesspiezoelectrictransducerswith nonuniformelectroelasticparameters. Dov HazonyandSustoR. Raya (Departmentof Electrical Engineeringand Applied Physics,Case Western Reserve University,Cleveland,OH 44106) RR3. Interpretingpiezoelectricoscillatordissipationeffectsin termsof the underlying materials viscoelastic relaxation phenomena. R. Lowell Smith (Texas Research Institute, 9063 Bee Caves Road, Austin, TX 78733) Ofcc.nccrnare the generationand detectionof soundby piezoelectric Equivalentcircuit modelinghas long beenrecognizedas a valuable transducers withnonuniform materialproperties alongtheprincipalaxis. It is shownthat subjectto conditionsof distortionless propagationIV. tool for describing transduction processes and the behaviorof trans- Burke,R. $. Duffin,and D. Hazony,"DistortionlessWavePropagationin Inhomogeneous Media andTransmission Lines,"Q. Appl. Math. 183194(July 1976)] it ispossible to associate a circuitmodelwiththetransducer similar to the Mason model [W. P. Mason, Electromechanical Transducersand WaveFilters (Van Nostrand, New York, 1948) ]. This work foillowsthe developmentof sphericalshellpiezoelectrictransducers [D. Hazony, "Nonuniformly PoledSphericalShell PiezoelectricTransducers,"J. Acoust.Soc.Am. 81, 1624-1627(1987)]. [Work supported in part by Tecsonics, Inc.] 1:50 RR2. Optimizing the performanceof piezoelectricdrivers usingstepped horns. A. Bangviwatand R. D. Finch (Department of Mechanical Engineering, Universityof Houston,Houston,TX 77004) ducers. Theformalism hasbeen generalized torepresent distribute. dproperties,modecoupling,andtheimplications of structuralcomplexity,mechanicalloading,and electricaltuning.Dissipationor lossphenomena canalsobe accommodated by equivalentcircuitmethods.It is generally recognized thatinterpreting all relevantelastic,dielectric,andpiezoelectric coefficients as complexnumbersis an appropriateway to represent losses.All toooften,the scopeof thisapproachdiscourages the development of detailedresults.This paperanalyzesthe dissipationmodeling problemin termsof relaxationprocesses, a phenomenological approach that givesalgebraicform to certaingroupings of the electromechanical coefficients. The implicationis that soundspeed,couplingcoefficients, materialsmoduli,etc., are frequencydependent.A mass-loaded longitudinalvibratorwith a segmented piezoelectric cylinderis analyzedasan exampleof thismodel-parameter dispersion problem.Of particularinterest is the effectof the adhesivejoints on the oscillatoradmittancespectrum overa wide rangeof frequencies. An analysisispresented for thedesignof piezoelectric transducers, in a sandwichconfigurationusinga steppedhornasparticlevelocityamplifier. The ca3ndition for maximumpowerdeliveryto the load impedanceis establishedin terms of the parametersof the horn and the piezoelectric material• Analytical solutionscan be found readily if one of the horn lengthsischosen.If thisisdone,thentherearetwo arearatiosthat satisfy the maximumpowercondition.The methodof falsepositionthat is suitablefor findingsolutionsnumericallyin a generalcaseby usinga comput- RR4. Developmentof sphericaland cylindrical polarization for threedimensionalpiezoelectricfinite elements.P. Tietee,'• K. Anifrani,'• J. er is aisc described. Fours Lea Plagcs,France) $95 d. Acoust.Sec. Am.Suppl.1, Vol.82, Fall 1987 2:20 C. Debus,J.C. Decarpigny(InstitutSup6rieurd'ElcctroniqueduNord, 41 Bd Vauban,59046 Lille Cedex,France), and D. Boucher(Groupe d'Etude et de Recherchecn Detection SousMarine, Le Brusc, 83140 Six 114thMeeting:AcousticalSocietyof America $95 Three-dimensional analysisof sphericalor cylindricaltransducers with axiallyor radiallypolarizedpiezoelectric ceramicsis ratherdifficult because it needsa largenumberof classical elementsto takeintoaccount the correctpolarization.A new type of piezoelectricfinite elementhas beendevelopedin the ATILA code [J. N. Decarpignyet al., J. Acoust. Soc.Am. 78, 1499( 1985) ] to obtaina goodrepresentation of the polarization.This paperdescribes theseelementsand analyzesthe results(in air resonance modes,electricalcharacteristics, and modeshapes)ob- of flow and soundsources.L. M. B.C. Campos(InstitutoSuperior tainedon someexamples. a) Currentlyat SINAPTEC, 41 Bd Vauban, T6cnico,1096LisbonCodex,Portugal) atedby thesourceagainstthecharacteristic impedance andtheotherby reflectingthe rearwardwave (incidenton the source)off the sourceimpedance.This principleis seminalin linkingthe two approaches. 3:20 RR8. On the natureof acousticsingularitiesarisingfrom the coincidence 59000 Lille, France. 2:35 RRS.Useof special waveforms foroptimum efficiency ofhigh-intensity soundsources. FredericG. Pla (SverdrupTechnology, Inc., NASA LcRCGroup,16530Commerce Court,P.O. Box30650,Midpark Branch,MiddleburgHeights,OH 44130)andGerhardReethof(Noise ControlLaboratory,157 HammondBuilding,UniversityPark, PA 16802) High-intensitysoundsources, suchassirens,havereceivedmuchat- tention intherecent pastduetorenewed interest inindustrial applications of high-intensity acoustics. Asa resultof theveryhighsound pressure levelsrequired( 155-165dB), finiteamplitudeeffectsmustbetakeninto accountin thedesignof soundgenerators. A timedomainsolutionof the second-order nonlinear waveequation isusedto predictthebehavior of initiallynonsinusoidal planewaves, andiscompared witha frequencydomainapproach. Results for initiallysinusoidal, rectangular, andin- Thecoincidence thatflowandacoustic sources canoccurin important acoustic problems, e.g.,a spherical wavein a conicalductcontaining an incompressible meanflow,isan exampleof flowandsoundsources colocatedat thevertexof thecone;theconeistheparticularcasen = 1 of the power-lawductsof cross-sectional areaS(x ) • x 2" whichexhibit,for all real n, coincidenceof flow and sound sourcesor sinksat x = 0 or x = It isshownthat thecoincidence of flow andsoundsourcescanlead,for the powerlaw ducts,to algebraicsingularitiesfor the phase,and essential singularities fortheacoustic field;a transformation involving thesolution of a particularRicattiequation isusedto accountfor theessential singularity, and regularizethe acousticproblem.Exactsolutionsare obtained for soundin parabolicandhyperbolic nozzles,andtheseareusedto study theacousticfieldin theray,asymptotic andcompactness approximations, compareacoustic velocityandpressure, discuss equipartition andbiasing of kineticand compression energies, and conservation and nonconserva- tion of waveaction.The dualityprinciplefor horns[R. W. Pyle, J. Acoust.Soc.Am.37, 1187A (1965)] isshownto fail,andhavenosimple extension,for nozzles. verse-shock wavesare presented. It is shownthat the shockformation distancefor an initiallyinverseshockwaveis•r timestheshockformation distancefor an initially sinusoidalwave, and that the wave distortion actually results inanamplification ofthefundamental, thusincreasing the efficiency of thesoundgeneration process. The consequences of wave distortionon severalpracticalhigh-intensity soundsources is discussed. 3:35 RR9.Whispering galleryresonances onsolidelasticspheroids. A. Nagl (Departmentof Physics,CatholicUniversity,Washington, DC 20064), M.F. Wcrby (NavalOceanR&DActivity, Code221,NSTLStation,MS 39529),H. Oberall(DavidW. TaylorNavalShipR&D Center, 2:50 RR6. The distortionspectrumof the fractionallyaddressed digital oscillator. W. M. Hartmann(Physics Department, MichiganState University,E. Lansing,MI 48824) Thetechnique offractional addressing permits onetoconstruct adigital oscillator with arbitrarilyhighfrequency resolution. However,the technique introduces distortion. Thedistortion powerspectrum maybe calculated exactly byborrowing mathematical methods fromcrystallography,suchthatthecrystallographic unitcelllengthequals thedenomi- Annapolis,MD 21402andDepartmentof Physics,CatholicUniversity, Washington,DC 20064), and J. W. Dickey (David W. Taylor Naval ShipR&D Center,Annapolis,MD 21402) UsingtbeNORDA T matrixcode,backscattering amplitudes versus frequency havebeencalculated fortheaxialincidence ofa planeacoustic wavein wateronsolidtungsten carbide spheroids witha largevarietyof aspectratios.Besides theusualseriesof broadRayleighwaveresonances, seriesof narrowwhispering galleryresonances areobserved andinvestigated.The dominantmodenumberof eachof theseresonances is identi- fiedbyaninterpretation of thebistatic scattering pattern.Forincreasing natorof theirreduciblefractionalpartof theaddress increment.The calculationexhibitsremarkablemathematicalsymmetries, whichlead to aspectratios,it is notedthat an increasing numberof the lowest-order whispering galleryresonances failtobeexcited.A possible explanation of simpleclosed-form expressions for thelevelsof thecomponents in the distortion spectrum andforthetotalrmsdistortion. Theexpressions show howthedistortion maybeminimized orpossibly employed asanalternativetodigitalFM inthesynthesis ofcomplex tones.[Worksupported by thisphenomenon isdiscussed, based onphase matching arguments forthe circumferential propagationof surfacewaves. 3:50 the NIH. ] 3:05 RR7. Analyticalmodelingof an activenoisecontrolsystemin a duet.M. L. MunjalandL. J. Eriksson(CorporateResearchDepartment,Nelson Industries,Inc., P. O. Box 600, Stoughton,WI 53589) Makinguseof the transfermatricesandelectroacoustic analogies, a standingwave analysisof the activenoisecontrolsystemin a duct is presented incorporatingthecharacteristics of the primarysource,aswell astheauxiliarysource.Analyticalexpressions havethusbeenderivedfor the ratioof the twosourcepressures andcertainotherratiosor relationsof interestfor completecancellation of noisedownstream of the auxiliary source.Theseexpressions are then comparedwith thoseobtainedfrom transferfunctions of thevariousblocksconstituting theblockdiagramof the entire system.In the process,equivalencehas been established betweenthe differententitiesinvolvedin the two approaches, one of whichisusedin acoustics andtheotherin systemtheory.In particular,it hasbeenshownthat the pressure generated by a sourceagainsta load impedance canbelookeduponasa sumof two pressure waves,onegener- S96 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 RR10. Soundradiation from beamsunder the action of movingline forces. R. F. Keltie and H. Peng (Department of Mechanical and AerospaceEngineering,North CarolinaStateUniversity,Raleigh,NC 27695-7910) The topicof soundradiationfrom beamsunder the actionof harmonic line forcesmovingat subsonicspeeds(M < 1) is studied.The nondimensionalsoundpowerexpressions are obtainedthroughthe integrationof the surfaceintensitydistributionover the entire beam.An asymptotic expression for the soundpowerin the low-frequency regionis derived dependingupon the characteristics of the fluid loadingand the spatial extentof the appliedforces.Numericalintegrations havebeenperformed to determinetheeffectsonthe radiatedsoundpowerof theMach number M, theacoustic lengthoflineforceKoL,andthewavenumber ratioy. The resultsshowthat: For beamsunderheavyfluid loading,the effectof the speedof themovingforceisnotpronounced; whilefor beamsunderlight fluidloading,the uniquecoincidence radiationpeakat ?'• 1 for a stationary force (M = 0) is split into two coincidencepeaks(locatedin the frequencyregion?'< 1 and ?'> 1, respectively)due to the effectsof the Dopplershift.The valuesofKoL that suppress the coincidence peaksare alsochangeddue to the motionof the line force. 114thMeeting:Acoustical Societyof America S96 THURSDAY AFTERNOON, 19 NOVEMBER 1987 TUTTLE SOUTH ROOM, 1:30 TO 5:00 P.M. SessionSS. Noise V and Bioresponseto Vibration II: Effects of Man-Made Noise on Animals and Animal Communication William C. Cummings,Chairman Oceanographic Consultants,5948 Eton Court, San Diego, California 92122 Chairman's Introduction--l:30 Invited Papers 1:35 SS1. Reviewof studieson the effectsof man-inducednoiseon marinemammalsof the Bering,Chukchi,and Beaufort Seasand how the resultshavebeenappliedto federal offshoreoil and gasmanagementdecisions. JcrryL.ImmandClevclandJ. Cowles (MincralsManagcmentService, Alaska OCSRegion, Anchorage, AK 99508-4302) The U.S. MineralsManagementServicehasbeenfundingandmanagingstudiesof the potentialeffectsof man-inducednoiseon marinemammals,particularlyendangeredcetaceans,in the Bering,Chukchi,and BeaufortSeassince1978,and hasexpendedapproximately$5,000,000towardthoseefforts.The purposeof suchstudiesis to provideinformationneededfor informeddecisionmakingpertainingto environmentally soundleasingand management of offshoreoil and gasdevelopment on the AlaskaOCS. Many of the noise/ marine mammalinteractionstudieshavebeenusedto establishor modify leaseterms or regulationsfor offshoreoperations in federalleaseareas.Also,theseresultshavebeenimportantin resolutionof litigation pertainingto OCSoil andgasleasing/exploration. Specificcaseexamples of howresultshavebeenappliedwill be presented andprojectionof futureAlaskainformationneedsin thisdisciplinewill be discussed. 2:00 SS2. Potential impacton seaturtles, dolphins,and fishesof explosivesusedin offshoreplatform removals. Edward F. Klima, Gregg R. Gitschlag, Maurice L. Renaud (National Marine Fisheries Service, 4700 AvenueU, Galveston, TX 77551-5997),andWilliamC. Cummings(Oceanographic Consultants, 5948Eton Court, SanDiego, CA 92122) A highincidence of strandings of seaturtles,bottlenose dolphins,andfisheswasrecorded onthebeaches in thenorthwestern Gulf of MexicoduringMarch-June1986,whenexplosives wereusedto removeseveral oil platformsin adjacentoffshorewaters.Prior to, during,and after March-Juneof the followingyear, 1987, strandings for thesameareaweresignificantly lower,or negligible. Recoverylocations of driftbottlesreleased at the siteof theexplosions werecorrelatedwith someof the strandings. Wild turtlesandbottlenose dolphins wereobserved in thevicinityof theplatformsduringremovaloperations. Abnormal(stunted)Kemp'sridley andloggerhead seaturtleswereplacedat variousdistances fromplatformsandexposed toexplosions. Preliminaryresultsfromexposure studies,includingthedeathof a Kemp'sridley50 yardsfrom theexplosionandan unconscious loggerhead turtle approximately1000yardsfrom the explosion,are described. Allowingfor inbottomloss,receivedsound-pressure levelswerecalculatedto be 235 dB (50 yards)and greaterthan200 dB wherea cagedloggerheadwasfound unconscious. 2:25 SS3. A review of noise effects studiesfrom the U.S. west coast. Dilworth W. Chamberlain (ARCO, Environmental Protection,515SouthFlowerStreet,LosAngeles,CA 90071) Concernoverthepossibility of effects fromgeophysical activityonmarineorganisms hasrecentlygeneratedseveralresearch studies in California.Resultsof effectsstudies by compressed air releases fromairguns (seismic energyreleases) onfishdispersion anduponphysicaleffects to fishhavebeenobtained. Thisinformation includesthe physicaleffectsof a simulated,towedgeoseismic array on northernanchovy(Engraulis mordax)eggs,larvae,andadults.Otherstudieshaveproduceddataabouttheschooling behaviorof rockfish (Sebastes spp.) exposed to airgunreleases. Additionalresearch addressing theeffects ofseismic energyreleases on Dungcncssc crab (Cancermagister)eggsand larvaeandon rockfishbehaviorarebeingplannedfor the watersof Californiaand Washington. 2:50 S$4. The influence of sound propagation conditions on the behavioral responseof whales to underwater industrial noise. Charles I. Maline and Paul R. Miles (Bolt Beranek and Newman Laboratories, Inc., 10 MoultonStreet,Cambridge,MA 02238) Resultsof recentstudiesinvolvinggray, humpback,and bowheadwhalesshowthat whalestendto avoid areaswith highunderwaternoiselevels.Data obtainedfromwhalebehavioralobservations duringcontrolled S97 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America $97 exposure to representative industrial noisesources permitdetermination oftheprobability ofavoidance Paof thesourceregionasa functionof thenoiselevelL,. The zoneof influenceof a sourcemaybedefinedasthe regionwherePa> 0.5.WhiletheL, requiredto produce thisdegree ofavoidance hasbeenfoundto dependon whalespecies andsourcetype,somegeneralizations maybemade.For low-frequency continuous noise,50% of thewhalesexposed havebeenobserved to avoidregionswheretheoverallL, ishigherthan 115to 125dB (re: 1 pPa). Soundtransmission conditions at a specific sitedetermine thedistance fromthesource,whereL, falls belowthePa = 0.5 criterionlevel.Thezoneof influence thushasbeenfoundto varyconsiderably for thetest sitesinvestigated. For example,a drillshipoperatingat a test site off the AlaskanBeaufortcoasthasan estimatedzoneof influenceradiusof 4 km, but, off the coastof California,the estimatedzoneof infiuenee radiusisreduced to I km. [Work sponsored by theMineralsManagement Service. ] 3:15 SS5. Reactionsof bowheadwhalesto drilling and dredgingnoise in the CanadianBeaufort Sea. W. JohnRichardson (LGL Ltd., Environmental Research Associates, P.O. Box280,King City, OntarioLOG IK0, Canada),BcrndWfirsig(MossLandingMarine Laboratories, P.O. Box 223, MossLanding,CA 95039), and CharlesR. Greene,Jr. (GreeneridgeSciences,Inc., 5276 Hollister Avenue,Suite408, Santa Barbara, CA 93111 ) Behavioral reactions ofbowhead whales, Balaenamysticetus, toseven30-to40-rainunderwater playbacks ofrecorded drillship anddredge noiseweredetermined in 1982-1984. Waterdepthswere10.-150 m withlittle or noice.Some(butnotall) bowheads oriented awaywhenreceived noiselevelsandspectra resembled those several kilometers fromactualdrillships anddredges. Duringsomeplayback tests, callratesdecreased, feeding ceased,andcyclesof surfacing,respiration,anddivingchanged.Sensitivityof variouswhalesto noisediffered. Roughly halfresponded toreceived noise levels ofabout115dBre:I pPabroadband, orabout110dBinone«octband(20-30 dBaboveambient).Suchlevelsoccurred3-11 km froma drillshipanddredgein theCanadian BeaufortSea.Bowheads occasionally wcrcseenlessthan5 km fromactualdrillships anddredges, where received noiselevelswereat leastashighasduringbriefplaybacks. Thussomebowheads mayhabituateto prolongednoiseexposure. Alternatively,onlythe least-sensitive individualsmayoccurlessthan5 km from drillships anddredges. [Work supported by U.S.MineralsManagement Service.] 3:40-3:45 Break ContributedPapers 3:45 SS6.Effectsofnoiseofa proposed oceanthermalenergyconversion plant 4:00 SS7. Drill-site sounds and bowhead whale calls. Charles R. Greene on marine animals--A preliminary estimate.Robert S. Gales and Suc (GreeneridgeSciences,Inc., 5276 Hollister Avenue,Suite 408, Santa E. Moore (SEACO, Inc., 2845-D Nimitz Boulevard,SanDiego,CA Barbara, CA 93111 ) 92106), William A. Friedl (Naval Ocean SystemsCenter, Kailua, Hawaii 96734), and JamesRucker (National Oceanographic and Atmospheric Administration, Washington, DC 20235) An analysisis presentedof the predictedunderwaternoisefrom a proposed 40-MW oceanthermalenergyconversion (OTEC) plantto be located approximately • mileoffshore ofKahcPoint,Oahu,Hawaii,and itspotentialeffects onthemarineanimalsof thearea.The requiredtotal warm and cold seawaterflow of approximately1.6 million liters/rain involves 16pumpsdrivenbya totalof 17000horsepower. Earlierstudies [C P. Janotaand D. E. Thompson,J. Acoust.Soc.Am. 74, 256-266 (1983)] calculatedthe noisegenerated by variousproposed OTEC systems,and comparedit with actual measurements on a I-MW research system(OTEC-1) operatingoff Hawaii. Theseanalyticprocedures are utilizedto predictthenoisefor the40-MW plant.Underwater noisegeneratedduringtheconstruction phaseisexpected to exceed theplantnoise, butfora relativelyshortperiod.Generalauditoryperception andexpectedresponses ofcetaceans andfishes oftheareatobothplantoperation and construction noisesarediscussed. The mostlikelybehavioral responses includetemporarydisplacement fromthe areaof the plantandpartial maskingof communicative signals.[Work supported by NOAA.] S98 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 Bowheadwhales,an endangered species,migratewestwardalong Alaska'snorth coastduring Septemberto October,when ice conditions are bestfor oil explorationson the outercontinentalshelf.Concernexists that marineactivitysoundsmay adverselyinfluencewhalebehavior.Un- derwatersoundsweremeasuredin the fall of 1986near two drillship operations20 and 35 km offshore,northwestof Kaktovik, Alaska,with waterdepthsof 34-35 m. Soundlevelsat range0.2 km, 20-1000 Hz, for threesupportshipsunderwayat 10-12 kn (standardspeed)were 130, 137, and 145 dB re: I 3tPa.The level from the strongestice-breaking source, samefrequency bandandrange,was148dB.Corresponding levels from the drillshipwere 135-136dB duringdrillingand cleaningand 130 dB during tripping.For measurements within 0.4 km, soundlevelswere highestin «-octbandsfrom50-100Hz. Hourlymeasurements for 20 days at 11 km from one site revealed a median level of 114 dB in the 20-1000- Hz band,with 5thand95thpercentilelevelsof 108and 125dB,respectively. Dominantsounds at thatdistance camefromthedrillshipandsupport ships.Continuousmonitoring during 20 days recorded206 bowhead calls,with 29 of themlocated9.5-24 km from the drillship.[Work supportedby ShellWesternE&P, Inc.] 114thMeeting:Acoustical Societyof America S98 4:15 perorgeese wasinitiatedontheAlaskaPeninsula andontheNorth Slope $$8. Usi[ng sounds tocontrolthemovements ofsesotters.R, W. Davis, F. W. Awbrey,andT. M. Williams(SeaWorldResearch Institute, Hubbs MarineResearch Center, 1700South Shores Road,SanDiego, CA ofAlaska. Theobjectives wereto:( l ) describe thebehavioral responses of geeseasan effectof noisefromaircraftoverflights;(2) recordandexam• incnoiseassociated withexperimental overflights of fixed-androtary- 92109) wingaircraftandspecific altitudes,airspeeds, andenvironmental condi- Fieldtestswereconducted in PrinceWilliamSound,Alaska,to determinewhethertheplaybackof artificialsounds,killer whalecalls,or sea otterpupcallscouldbeusedto controlthemovements of seaotters(En- tions; (3) securebaselinenoise levels for routine flights of nonexperimental aircraft;and (4) providerecommendations to governmentagencies forreduction ormitigation ofanyadverse impacts associatedwithaircraftnoisetothesepopulations ofgeese. Behavior ofgeese was monitored fromremotelocations priorto, during,andfollowinglevel hydralutris).Testswereconducted on threegroups of fiveadultotters thatwerecaptured andplaced in anenclosed lagoon( 180m longX 40m wide).Sounds werebroadcast in air andunderwater froma platformin thecenterof thelagoon.The behavioranddistanceof theottersfrom the platformweredetermined fromvideorecordings obtainedfroma remote- overflights alongprescribed routes. Concurrently, «-octsound exposure and maximumA-weightedand soundlevelsof eachaircraft weremeasuredwith a realtimeanalyzer.Here,thedesignandpreliminaryresults of the firstof a multiyearstudywill bediscussed. controllext videocameraoverlooking thelagoon.Theresults showed that seaotterswerenot repelledby loud (SPL = 120dB at 1 m), obnoxious soundsin air suchasa warblelone (frequency-modulated sinusoidcen- teredat ItkHz) orair hornsandnarrow-band pulses at frequencies orS00 Hz-20 kHz projectedunderwater.However,the otterswereattractedto theaerialplaybackof seaotterpupcallsandrepelledby the aerialand underwater playback of killerwhalecalls.Psyehoacoustic stimuliappear to havethebestchanceof influencing themovements of wildseaotters. [Thisresearch wassupported by MineralsManagement Service, Contract No. 14-12-0001-30256.] 4:45 SS10. Acousticcommunication in P. Tigris.Harry Hollien (IASCP, Universityof Florida, Gainesville,FL 32611) Whilenootherspecies exhibitstheextensive language systems of humans,a numberare seento passinformationby oneor morebehaviors. Prominentamongthe observedsignalsare thoseof an acousticnature; 4:30 SS9. Effectsof aircraftnoiseon Pacificblackbrantandothergeesein Alaaka. David H. Ward, Dirk V. Derksen (U.S. Fish and Wildlife Service,1011 East Tudor Road, Anchorage,AK 99503), and Paul D. Schomer (U.S. Army ConstructionEngineeringResearch Laboratory,P. O. Box4005,Champaign, IL 61820) In 1987,a studyof the effectsof aircraft noiseon the behavior,distri- bution,zaadhabitatuseof Pacificblackbrant,Canadiangeese,andera- indeed, somesubspecies ofbirdsdemonstrate fairlycomplex communicativenetworks. AmongthePanthera, P. Tigrisisperhaps themostvocal. ExtensiveS/R experimentation with sevencaptivetigersandsubjective observation of nineothers(plusevaluationof a fewcallsrecordedin the field), haveresultedin classification of ten meaningful utterances plus two othersthat maybe informationbeating(not yet "decoded").This systemappearsto be somewhatmorecomplexthan that exhibitedby Tursiops T. (and certainlymoresothanthatof thecommondolphin). Thesecond stageof thisprojectinvolves playbackofrecorded andsynthesizedcalls in order to confirm (or not) the behaviorsnoted. THURSDAY AFTERNOON, 19 NOVEMBER 1987 TUTTLE NORTH ROOM, 1:30TO 5:00 P.M. SessionTT. Structural Acoustics and Vibration V: Viscoelastic Materials: Their Characterization and Application Fred Sehloss,Chairman Wilcoxen Research, 2096 GaitherRoad,Rockville,Maryland20850 Chairman's Introduction--l:30 1:35-2:00 TT1. PanelDiscussion: Characterization of Viscoelastic Polymers Wayne T. Reader, Moderator PANEL MEMBERS: B. L. Burkewitz R. J. Capps B. Hartmann R. F. Landel L. Rogers D. J. Townend R. J. Wetton S99 d. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 114th Meeting:AcousticalSocietyof America ContributedPapers 2:00 TT2. Structure and acousticalpropertiesof PTMEG polynrethanes. Corley M. Thompson (Underwater Sound ReferenceDetachment, occurin practice,the theorydoesnot permit mastercurvesto be constructedby theusualWLF frequency shiftmethod.Implicationsof this fact for dealingwith experimentalbulk modulusdata are discussed. Naval ResearchLaboratory,P.O. Box568337,Orlando,FL 32856-8337) and Karen L. Gebert (PCR, Inc., P.O. Box 1466, Gainesville, FL 2:45 328602) Polyurethanesare widely usedin underwateracousticaldevicesbecauseof their convenience and attractivephysicalproperties.Rarely, however,havethesepolyurethanes beendesigned for thespecificapplication. More typically,a commercialmaterialis usedin both sound-transmissionand sound-absorption applications,usuallywith little under- standingof structuralparameters thatcontrolthe acoustical properties. Thispaperwill showthattheacoustical properties ofthesepolynrethanes maybemodifiedby changing thesizeof thePTMEG softblock, andthat adjustment of thesoftblock components is ineffective in producing such changes. Evidencewill bepresented that thestructuralmechanism for the correlationbetweenacousticalpropertiesand softblocksize is the decreasingsolubilityof the hardblockmaterialwith increasing softblock molecularweight.Thisdecreased solubilityresultsin a morepuresoftblockandin lowerdynamiclosses andlowerspecific acoustic impedance. The evidence for thisistheregularchangein thesoftblock's glasstransition temperaturewith molecularweight.The acousticalramifications of theserelationships will be discussed. TTS. An algorithmfor applyingtheWilliams,Landel,andFerrymethod of reducedvariables. JamesJ. Dlubac,RichardJ. Deigan,andWayne T. Reader (Code 1905.2,Ship AcousticsDepartment,David Taylor Naval ShipR & D Center,Bethesda,MD 20084-5000) The measurement of the dynamiccomplexmodulusof viscoelastic materials onanyapparatus is confined to a limitedfrequency andtemperaturerange.Williams,Landel,andFerry(WLF) [J. D. Ferry,g'iscoelastic Properties ofPolymers ( Wiley,NewYork,1980), 3rded.Chap.11] reasoned thatan increase in temperature produces an identicaleffectin a viseuelastie materialasa decrease infrequency andviceversa.Empirically, then,thereis a correspondence between temperature andfrequency. The WLF methodof reducedvariableswasproposed as a methodof extending theeffective frequency rangeof thedynamicmodulusandloss factor.Thistechnique involves measuring thecomplex modulus overa limitedfrequency rangeanda varietyof temperatures. Thedataat differenttemperatures arethenplottedversus frequency andshifted alongthe frequencyaxisuntil a smoothcurveresults.The effectsareshiftedmodu- luscurves describing thecomplex modulus overa broadfrequency range 2:15 Tr3. Relationship between italyurethane compositionand viscoelastic propertiesof model urethanesystems.R. N. Capps, G. M. Stack, E. M. Dodd (Underwater Sound Reference Detachment, Naval Research Laboratory, P.O. Box 568337, Orlando, FL 32856-8337), and E. Y. Chang (American Cyanamid,Bridgewater,NJ 08807) A studywasperformedto determinethe relationshipbetweenpolyurethanecompositionand viscoelastic propertiesin a seriesof model urethanecompounds.The compositionalvariablesstudiedincludedthe natureof the softsegment,amountof hard segment,and cure stochioroetry.Model urethanes werepreparedusingthealiphaticdiisocyanate, metatetramethyleue diisocyanate(m-TMXDI). This aliphaticisocyanateis lessreactivethancommonlyusedaromaticisocyanates anddoesnot undergothesecondary crom-linking reactions commonlyobserved with aromaticisocyanates. The urethaneschosenfor thisstudycontainedsoft segments basedon hydroxy-terminated polybotadiene (HTPBD) or polycaprolactone (PCL), whichis a highlypolar materialin whichextensive phasemixing shouldoccur. In contrast,polyurcthanes basedon HTPBD shouldundergomorecompletephaseseparation. Prcpolymers werepreparedwith both of thesesoftsegmentsand with availableisocyanatecontentsvarying from 3.7% to 5.7%. Diethyltoluenedismine {DETDA) wasusedto cure theseprepolymersover a rangeof cure stochiometries.Therma]transitionsattributedto the hard and softsegments wereobserved in theseurethanes usinga Perkin-ElmerDSC-4. Dynamic meehancial behaviorwasmeasured usingbotha PolymerLabsdynamic mechanical thermalanalyzer(DMTA) anda resonance apparatus developedat theNRL-USRD. [Sponsored by ONT.] 2:30 at somefixedtemperature. Typically,theWLF shiftrequires several runs at temperatures closeenough sothatthemodulus ofonerunoverlaps the next.In thispaper,theWLF methodof reducedvariables isreviewedand analgorithm foraccomplishing theWLF shiftisinvestigated. Thisalgorithmshiftsdynamicmodulus dataat varioustemperatures, theconsecutiverunsof whichmaynotcontainoverlapping toodull. 3:00 TY6. Effects of aging on dynamic bulk moduli of several elastomers. J. Burns (Florida Instituteof Technology,Melbourne,FL 32901), P. S. Dubbelday, and R. Y. Ting (Underwater Sound Reference Detachment, Naval ResearchLaboratory, P. O. BOx 568337, Orlando, FL 32856-8337) Samples of naturalrubber,butylrubber,neoprene, andpolyurethane wereagedfor a total of 24 monthsat roomtemperaturein seawater.The dynamicbulk moduluswasmeasuredfor eachsampleat intervalsduring the2-yr agingperiod.Age-induced changes in thedynamicbulk moduli wererelativelysmallin all cases, butthereweredistinctdifferences among the varioustypesof elastomerstested.In general,agingeffectsappear considerably more pronouncedfor shearand Young'stoodullthan for bulk toodull. Possible reasons for this are discussed. 3:15 TY7. An automatedsystemfor characterizingthe vl'brationdamping propertiesof mater'als.M. L. Drake (Universityof Dayton research Institute,300 CollegePark, Dayton,OH 45469), R. G. Smiley(Entek Scientific Corporation, Cincinnati, OH), and D. M. Hopkins ( Universityof Dayton ResearchInstitute,300 CollegePark, Dayton, OH 45469) •F4. Bulk modulusthermoviseoelastieitytheory for rubberyelastomers. J. Burns(Florida Instituteof Technology,Melbourne,FL 32903) and P. S. Duõbelday (Underwater Sound Reference Detachment, Naval ResearchLaboratory,P.O. BOx568337,Orlando,FL 32856-8337) Earlierattemptsto developa free-volumetheoryof thedynamicbulk modulus ofrubberyelustomers [J. Burnseta!.,J.Aeoust.Soc.Am. Suppl. I 79, S62 (1986) ] led to separateexpressions for the frequencydependenceof the modulusat constanttemperatureand for the temperature dependence atconstantfrequency. Thesehavenowbeencombined togive a free-volume-based theoryfor the bulk modulusas a functionof both frequencyand temperature.For certainrangesof parameters that may S100 J. Acoust.Soc.Am.Suppl.1, VoL82, Fall1987 In 1980 ASTM finalizedthe dampingcharacterizationstandardE756-80 around the resonantbeam test method (M. L. Drake and G. E. Terborg, TechnicalReport AFWAL-TR-80-4093, 15 January 1976-31 December1979). The Universityof Dayton, in conjunctionwith Entek Scientific,developeda computer-aided testsystemthat automatedthe test procedures and data reductionequationprescribedin E-756. This softwarepackage,combinedwith theappropriatehardwaresystem,resultsin an efficientworkstation for conducting dampingcharacterization studies of materials anddesigning effective dampingsystems utilizingthedampingproperties determined. The presentation will outlinethe philosophy and operationof the system,includingdiscussions on the systemauton- 114thMeeting: Acoustical Societyof America S100 omyandthefunction ofeachsystem component; atypicaltestsetup, data reduction, andhomogram generation; andanexample of a damping designdevelopment. example,thesystemmaybeusedto investigate thedynamicproperties of viscoelastic polymerscontainingfinite sizedinclusionssuchas chopped fibersor plateletfillers.Exampleswill be presentedof the system'suseto examine thereinforcement provided byglassfibersofseveral lengths ran3:30 domlydispersed throughouta viscoelastic polymer. TYS. Dynamic material property measurements using the Polymer Laboraturies' dynamic mechanical thermalanalyzer.RichardJ. Dcigan, WayneT. Reader,and JamesJ. Dlubac (Code 1905.2,ShipAcoustics Department,David Taylor Naval Ship R & D Center, Bethesda,MD 20084-5000) Thebasicprincipleof measurement of anydynamicmechanical property apparatusis to infer the elasticand lossmodulifrom the measured responseof a sampleand a known theoreticalsolution.The extentto whichthemeasured response ismodeledby thetheoreticalsolutiondeter- minesthevalidityof theinferredmaterialproperties. Thepurpose of this paperis to investigate the measurement of the dynamictoodullof viseoelosticmaterialsusingthedynamicmechanical thermalanalyzermanufactured by Polymer Laboratories,Inc. The complexYoung's modulus [E * = E'( 1 + i•/E) ] or shearmodulus[G * = G' ( 1 + i•la ) ] is determinedat frequencies rangingfrom0.01to200Hz at temperatures ranging from -- 150'C to 300'C. The Young'smodulusis foundthroughthe bendingof eithera doubleor singlecantileverbeam.The WLF shiftis employedon data at severaldifferenttemperatures to generateshifted modulus curvesofE' and•/Eand/orG' and•/a versus extended frequency for fixedtemperature. Computer-generated shiftedmoduluscurvesofviscodastic materials are discussed. 3:45 'Fr9. Viscoelastic constants for pressurereleasematerials.Gary Caille, Jacek Jarzynski,Peter Rogers,and GeorgeW. Woodruff (Schoolof MechanicalEngineering,GeorgiaInstituteof Technology,Atlanta, GA 4:15 TTll. Constrainedviscoelasticlayer asusedto dampenstructuresunder multiplestrainingmodes.M. Bonnetand B. Garnier (Metravib R.D.S., 64 ChemindesMouilles,69130Ecully,France) The designof viscoelastic dampingmechanisms requiresbotha thoroughknowledgeof the viscoelastic propertiesof materialsand proper dimensioning andtoolscapableof handlingall cases.This is particularly importantwhendealingwithaerospace structures wheretheweight-toperformance ratioiscritical.Thispointis illustrated usinga specific example.Here,analuminumhoneycomb/carbon skinsandwich isdamped to handleexplosion shockwavesovera frequency bandwidthof 1 to 10 kHz at a temperature of -- 20 øC.Dynamicmechanical analysis(in modulusand lossangle) on a varietyof materialsled to the selectionof an appropriatepolyurethaneelastomer.The resultingsandwichstructure accounts not only for bendingmodes(usingRuzicka'stheory),but for tensioncompression andshearmodesaswell (inspiredfromY.V.K. SadasivaRaoandB.C. Nakra). Thiscompromise is obtainedthroughthe development of specificsoftware,whichisbothsimplerandquickerthan a finiteelementanalysis.A goodagreement betweentheoryandexperimentat roomtemperature isshown.Thewaterfallvisualization technique usedisparticularlyusefulin detailingapparentmechanisms forall configurations: bending,shearing, andtensioncompression. Theseresultsareso encouraging that furtherdevelopment of thistechniqueisjustified. 30332) 4:30 A comparison of experimentally andtheoreticallydeterminedviscoelasticconstants for closedcellneoprene rubberandcorpreneispresented. Both of thesematerialshaveair concentrations of greaterthan 50% by volumeand exhibitlargeacousticattenuation.The longitudinalsound speedis approximately150m/s for the closedcell neoprenerubberand 340m/s for thecorprene.The experimental constants determinedarethe complexYoung'smodulusand complexplane-wavemodulusin the frequencyrangeof 500 to 5000Hz. Standardexperimental techniques for measurement ofthesetoodullareused.The experimental dataarecorrected for multipleinternalreflections by useof thecomplexcepstrumsignal processing technique. The theoreticalpredictions are basedon multiple scattering modelsandquasistatic meanelasticfieldmodels.[ Work supported,in part, by ONR Code 1125A.] 4:00 TYI0. Dynamic shear modulus measurements of fiber reinforced Imlymersusingforcedtorsionalvibrations.WayneT. Reader,James J. Dlubac, and Richard J. Deigan (Code 1905.2, Ship Acoustics Department,David Taylor Naval Ship R & D Center,Bethesda,MD 20084) TT12. Measurementof complexelectromechanicalpro•rties of highly damped piezoelectric materials. F. DouglassShields (Physics Department,Universityof Mississippi,University,MS 38677) and Kurt M. Rittenmyer (Underwater Sound ReferenceDetachment, P.O. Box 568337, Orlando, FL 32856-8337) The mechanicalresonance techniquefor determiningthe dielectric, elastic,and electromechanical propertiesof piezoelectricmaterialshas beendevelopedand refinedfor over40 years,asdescribedin the different versionsof the IRE-1EEE standardon piezoelectricity. The conventional methodassumes the piezoelectric resonatoris nearlymechanicallylossless.This assumptionis valid for somepiezoelectricceramiccompositions,but is inaccuratefor morehighlydampedmaterialsusedin hydrophoneapplications. New methodsbasedon the measurement of the real and imaginaryparts of the complexadmittancehave beendevisedby Sinits [IEEE Trans. SonicsUltrason. SU-23, 393-402 (1976)] and Sai- tah et al. [IEEE Ultrason.Symp.Proc. 620-623 (1985) ]. One of these methodshasbeenacceptedand the parameterfitting routinesimproved uponto determinethe electromechanical propertiesof severalimportant piezoelectric materialsthat havea high degreeof damping.The importance,applications, and limitationsof this methodwill be describedfor thesepiezoelectricmaterials. A forcedtorsionalvibrationsystem developed by theNationalBureau of Standards[E. B. Magrab, J. Res. Natl. Bur. Stand.89, 193-207 (1984)] for the DavidTaylorNavalShipResearch and Development Centeris usedto measurethe dynamicshearmodulusof viscoelastic ma- terials.Thecomplex shearmodulus G * = G '( 1 + bla ) isdetermined by measuring therelativeamplitudeandphaseat theendsof a harmonically torquedrightcircularcylinder.Theshearstoragemodulus andlossfactor arefoundat frequencies rangingfrom 50 to 1000Hz for fixedtemperaturesrangingfrom -- 40 øCto 70 øC.An algorithmbasedupontheWilliams-Landel-Ferrymethodof reducedvariables[J. J. Dlubac et al., J. Acoust.Soc.Am. Suppl.1 82. (1987), precedingabstract]is appliedto theisothermal datasetsto generate curvesof G' and•/Gversusextended frequency rangefor a fixedtemperature. A majorfeatureofthesystem is thatthecylindrical specimen dimensions mayrangeupto 10cmindiameter and 15 cm in height.The advantage,of course,is that the effective complexmoduli of inhomogeneous materialsmay be determined.For S101 d. Acoust.Soc. Am. Suppl. 1, Vol. 82, Fall 1987 4:45 •r13. Surface waves in viscoelastic fluid. H. W. Jones, H. W. Kwan (Department of EngineeringPhysics,Technical University of Nova Scotia, Halifax, Nova Scotia B3J 2X4, Canada), and E. Yearman (Department of Electrical Engineering, Imperial College, London, United Kingdom) Viscoelastic fluidswill, in principle,propagateLamb and Rayleigh waves.In thesecircumstances, the shearmodulusis frequencydependent in both its real and imaginaryparts;therefore,it mightbe expectedthat someunusualeffectswill occur.In this paper,the propagationof such wavesby referenceto the viscoelastic propertiesof crudeoil is described. Numerical calculationsshowingthe propertiesof thesewavesunder a rangeof conditions arepresented. 114th Meeting:AcousticalSocietyof America S101 THURSDAY AFTERNOON, 19 NOVEMBER 1987 UMS 4, 1:30 TO 4:20 P.M. SessionUU. Underwater AcousticsVII: PropagationModeling Henrik Schmidt, Chairman Departmentof OceanEngineering,Massachusetts Instituteof Technology, Cambridge,Massachusetts 02139 Chairman's Introductinn--l:30 ContributedPapers 1:35 UUI. Numerical modelingtechniquesin underwater acoustics.Paul C. Etter (ODSI DefenseSystems,Inc., 6110 ExecutiveBoulevard, Rockville, MD 20852) A summary of available underwater acoustic modeling techniques is presented, emphasizing thosecomputer-implemented codesapplicable principallyto research in supportof navalsonardesignandoperation. This comprehensive summaryaddresses threemajorclasses of models: propagation loss,noise,andreverberation. Fifty propagation lossmodels arecategorized according tofivetheoretical derivatives ofthewaveequa- the orthogonal (u,v) observationplane the caustic has the form D(u - uc)3= v2.Thepresent research isconcerned withthecaustic surfacegenerated by considering thez dependence of thecausticparameters D and uc. Reflectionand transmission problemsthat lead to this W(x,y) in theparaxialapproximation arediscussed aswell asa novelpropertyof mergingrays.Soundradiatedby a point sourceso as to reflectfrom a surface whose height relative to the xy plane is of the form h•x2+ h•y•x + h3.V2produces a transverse cusp. Thereflected wavefield is describedby the Pearceyfunction.Transversecuspsweresimulatedby reflectinglight from a surfaceof thisform. [Work supportedby ONR. ] tion:raytheory,multipath expansion, normalmode,fastfield,andparabolicapproximation; furtherdistinctions aremadebetween range-independentand range-dependent solutions.Fourteen noise models are dividedintothe categories of ambientnoiseandbeam-noise statistics; models in thelattercategory arefurthersegregated according to analytic and simulationapproaches. Fourteenreverberation/active sonarmodels arejointlydescribed in additionto twomultiple-component modeloperatingsystems; the reverberation modelsaredistinguished according to cellandpointscattering formulations. Supporting open-literature citationsareprovidedto identifyprincipalinvestigators andinstitutions. This summaryaffordsa convenientbaselineagainstwhichto definefuture developmental effortsin suchareasashybrid,3-D, broadband, stochastic, shallow-water, andunder-ice acoustic modeling. 1:50 UU2. Derivation of higher-order parabolic equations which include deasityvariations.GeorgcH. Knightly andDonald F. St. Mary(Center for AppliedMathematicsandMathematicalComputation,Department 2:20 UU4. Resultsof a very-low-frequency acousticpropagation experiment in the Cascadia Basin. Hassan B. Ali, Craig Fisher, Mona Authcment(NavalOceanResearch andDevelopment Activity,NSTL, MS 39529-5004),andJeff Beckleheimer (ODSI DefenseSystems, Inc., 6110 ExecutiveBoulevard,Rockville,MD 20852) A series of very-low-frequency (VLF) measurements wereconducted recentlyby the Naval OceanResearchand Development Activity (NORDA) in theCascadia Basinoff theOregoncoast.Usinga fixed verticalarrayof 16hydrophones anda distribution of oceanbottomseistoometers(OBS), the responses to cw, explosive, and air gun sound sources weremonitored. Thepartitioning ofacoustic energy between the waterborne andbottompathsisexamined, usingtherelativeresponses of OBSandwater-column hydrophones. Comparisons between theexperimentalresultsand predictions basedon standardnumericalmodelsof acoustic propagation areusedtoascertain theeffects onVLF propagation ofbottomshearandrange-dependent topography. of Mathematics,Universityof Massachusetts, Amherst,MA 01003) Severalapproachesto the derivationof parabolicequationsare explored,whichincludegeneraldensityfunctions,startingfrom the acoustic waveequationin two andthreedimensions. In particular,higher-order (wideangle)parabolicequations arepresented. Alsopresented arefinitedifferencediscretizations of the equationsthat ( 1) attemptto alleviatethe needfor specialconsideration of the "continuityconditions"whenone fluid mediuminterfaceswith another,and (2) specificallytake into considerationsuchinterfaces.Mathematicalcomparisons are madebetween thesetwo approaches. 2:05 UU3. Transversecuspcausticsproducedby reflectionand transmission: The causticsurfaceandopticalsimulations. Philip L. MarstonandCarl K. Frederickson (Departmentof Physics, Washington StateUniversity, Pullman, WA 99164) High-frequency soundreflectedfromcurvedsurfaces or refractedby inhomogeneities mayproducecusped caustics thatopenuproughlytransversetothedirection ofpropagation. Thoughthebetter-known longitudinalcuspor areteis produced by a con9ave cylindricalwavefront,trans- 2:35 UUS. High-speed normal mode calculations via Milne's equation. F. J. Ryan (Code541,NavalOccanSystems Center,SanDiego,CA 921525OOO) A majorimpediment to theuseofnormalmodemethodsindeepwater and/or at high frequencies is the calculationaltime per mode.A fast, numericallystablemethodof modecalculationisdescribedwhichisbased uponMilne'snumericalsolutionof theSchr6dinger equation.The gistof Milne'smethodisa phase-amplitude transformation of thelinearsecondorder ODE for the modal depthfunctionsinto a nonlinearODE for the quantum momentum. The resultingnonlinear equation hasimproved nu- mericalstability,particularlyfor high-ordermodes.In the limit of large wavenumber,the method reducesto the conventionalJWKB form, but, unlikeJWKB, it isvalidat turningpoints.An accurateandefficientcomputationalmethodis developedthat yieldssecond-order convergence to aneigenvalue. Thetechnique isdemonstrated forsingleandmultipleduct sound-speed profilesand comparedto conventionalpropagatormatrix approachesfor modalcalculations. 2:50 versecuspsare producedby distinctlynoncylindricalwavefrontsof the formW(x,y) = -- (a•x2q-az•x q-a3•), wherea2•0andthedistancez totheobservation plane• ( - 2a•) - • [Marston,J.Acoust.Soc.Am. 81, UU6. An efficientenhancement of finite-differenceimplementations for solvingparabolicequations.$. S. Robertson,D.C. Arney (Department 226-232 (1987)]. The wavefront propagatesin the z directionand in of Mathematics, U.S.M.A., $102 J. Acoust.Soc. Am. Suppl.1, Vol. 82, Fall 1987 West Point, NY 10996-1786), 114th Meeting:AcousticalSocietyof America W. S102 L. Siegrnann,and M. J. Jacobson(Department of Mathematical Sciences, Rensselaer Polytechnic Institute,Troy, NY 12180-3590) The parabolicapproximation methodis widelyrecognized asuseful for accuratelyanalyzingsoundtransmissions in diverseoceanenvironments. One reason for its attractiveness is because solutions are marched in range:, therebyavoidingthe massive internalstorage requiredwhen usingthe full waveequation.Presentimplementations employa range stepsizethat is prescribed eitherby theuseror by thecodeandremains fixedfor'theduration of thecomputation. An algorithm ispresented in whichthe rangestepis adaptivelyselectedby a procedurewithin the implicitfinite-difference (IFD) implementation oftheparabolic approximation.An errorindicatoriscomputed ateachrangestep,anditsvalueis comparedto a user-specified error tolerancewindow.If the error indica- torfallsoutside thiswindow,a newrangestepsizeiscomputed andused untilthe.errorindicatoragainleavesthetolerance window.For a given tolerance,the algorithmgenerates a rangestepsizethat is optimalin a specified. sense andwhichoftenleadsto largedecreases in run time.Additionalrelatedmodifications to the IFD implementation will alsobe discussed. Severalexamplesarepresented that illustratethe efficacyof the enhanced algorithm.[Work supported by ONR.] A numericallyefficientprocedurefor findingcompletesolutionsto wavepropagation problems in elasticmediawhoseparameters dependon depthonly,butwhereoneof theinterfaces (e.g.,water/air) isallowedto vary (undersomerestrictions)with range,is derivedfrom the method recentlyproposed byH. SchmidtandF. Jensen. Thisextension applies to thecasewherethedepthataninterface expressed asa function ofrangeis a sumofsinusolds. Theroughness imposes energy coupling equations that mustbesatisfied bysetsofwavenumber components of thepressure field, groupedaccording to thewavelengths ofthesinusolds. Theimportance of thecouplingdepends ontheamplitudes of thesinusoids, andaccounts for al nonevanescent forwardandbackscattering. Plane-wave reflectioncoef- ficientcalculations usingthismethodagreewith exactcomputations basedon work by R. Holford [J. Acoust. Soc. Am. 70, 1116-1128 ( 1981) ]. Thisprocedure isfurtherextended toallowtheroughinterface, say, the free surface,to move (under somerestriction) with time, thus allowinga rapidcompletesolutionof the waveequationincludingfull Doppler effects. 3:50 3:05 UU7. Etfectsof phaseerrorson pulsesynthesis.David H. Wood (Code 3332,NavalUnderwater Systems Center,NewLondonLaboratory, New London CT 06320) and Robert P. Gilbert (Center for Scientific Computation,Universityof Delaware,Newark, DE 19716) UU10. Roughsurfacescatteringusingtheparabolicwaveequation.Eric I. Thorsos(AppliedPhysicsLaboratory,Collegeof OceanandFishery Sciences, Universityof Washington, Seattle,WA 98195) The accuracyof the parabolicwaveequation(PE) for calculating scatteringfromrandomlyrough,pressure-release surfaceshasbeenexaminedthroughcomparison withexactnumericalresultsbasedonsolvingan One way ofmodeling pulse propagation i• theocean istodecomposeintegral equation.To facilitate comparisonwith exact results,the PE the pulseinto a weightedsumof functions ei•', summedovervarious values ofto.Onethen•olvestheHelmholtzequation numerically foreach requiredvalueof to.The weightedsumof thesefieldsthengivesthe desiredresponse to thepulsein question. An alternateapproach solves the parabolicequationfor eachrequiredvalueof to,ratherthan the Hclmholtzequation,therebyintroducingphaseerrors.Thesetworesultscanbe relatedbyanintegraltransform, or transmutation, thatconverts theapproximatesolutionobtainedfrom the parabolicequationinto the true solutionobtained fromtheHelmholtzequation. As a result,corrections can be made and/or errors can be estimated or bounded. problemisalsoconverted to oneof solvingan integralequation:the"PE integralequation."For the casesexaminedto date,the PE accurately predicts lowgrazing angle(0s<20ø)forwardscattering, including multiplescattering andshadowing effects. As thegrazingangleincreases, it is foundthat themainerrorisan angularshiftarisingfrompropagation to and from the surface,not from the surfacescatteringitself.Finally, a methodhasbeendevisedto computelow grazinganglebackscattering with the PE. This requiresthe sequential solutionof two integralequationsto accountfor multiplescatteringin boththeforwardandbackward directions.[Work supportedby ONR.] 3:20 UUS. RigorousGaussian-beam modelingof sourcefunctionsfor•coustic radiation and propagation.John J. Maciel (Missile SystemsDivision/ RadomeSection,RaytheonCo., Bedford, MA 01730) and Leopold B. Felsen( DepartmentofElectricalEngineering andComputerScience, PolytechnicUniversity,Farmingdale,NY 11735) Gaussianbeamfunctionshavefavorablepropertiesfor propagating acousticwavefieldsthroughcomplicatedoceanenvironments.Use of thesefunctionsin propagation algorithmsrequiresa decomposition of the actualor inducedsourcedistributioninto a superposition of Gaussian beams.In implementationsof the "Gaussianbeam method" so far, the decomposition isnot uniquebecause of freelyassignable parametersin the beamstack.This causes difficultieswith a priori predictability[ Lu et al., Geophys.J. R. Astron.Soc.(1987); Niver et al., J. Acoust.Soc.Am. Suppl. I 81, S9 (1987)]. The sourcerepresentation problemcan be addressed rigorouslyby performingthe decomposition on a latticein a discretized(configuration-spatial wavenumber)phasespace.The formulationof thisdiscretization scheme[Bastiaans, Opt. Eng.20, 594 ( 1981) ] is reviewedand then appliedto radiationfrom a cosineaperturetestfield distribution.It is shownhow successive additionof individualdisplaced and(or) rotatedbeamswith narrow, wide, or "matched"waistssystematicallyhomesin on the independentlycalculatedreferencesolution,althougheachselectionstrategyemphasizes differentregionsin the phase space.Tlaeutility of the variousoptionsis discussed, as are the implicationsfor synthesis of aperturefieldsfrom measuredfarfielddata. 3:35 UU9. A full wnve solution for propagationin horizontally stratilied elastic media with full forward and backscatter.Ziad Haddad (AT&T Bell Laboratories,14A420,WhippanyRoad,Whippany,NJ 07981) S103 J. Acoust.Soc. Am. Suppl. 1, Vol. 82, Fall 1087 4:05 UUll. Multiple array processors for three-dimensional •coustic problems. Yu-chiung Teng (Aidridge Laboratory of Geophysics,ColumbiaUniversity,New York, NY 10027) Applied Presently available analytictechniques for solvingwavepropagation problems areonlyusefulforsimplecases. In realisticcases, thepresence of inhomogeneities andirregularboundaryconditions defiesanalyticalsolutions.Oneofthebestnumericaltechniques suitable forsolvingwavepropagationproblemsin a complexgeologicalmediumis the finite element method.In principle,thefiniteelementmethodiscapableof dealingwith any two- or three-dimensional acousticand elastodynamic problems. However,the computerincorestorageand computingtime for solutions increasegreatlywith eachdimension andare beyondthe capabilities of theconventional computers. It is,therefore, alwaysdesirable tosearchfor alternatives that mayreducethecomputational labor.In takingadvantageof thecurrentadvances in hardwaretechnology, thegoalisto develop the finiteelementalgorithmfor an extraordinarily fastand inexpensive computersystem.In particular,the implementation of the finiteelement methodon multi-arrayprocessor systemwasconsidered. The algorithm, basedonthenodal-podal-point-oriented approach, for implementing the finiteelementmethodon theparallelcomputation systemhasbeendeveloped.As a demonstrative example,the three-dimensional problemfor the acousticcasehasbeensuccessfully testedon the IBM looselyCoupled Array Processors (ICAP) by usingoneto eightprocessors. By usingthe 31X 31X 3l-elementmodelon PRIME 750, the computingtime for 200 time stepsis about 8.8 h and I/O time is 1.7 h. If usingthe ICAPI to calculatethe samemodel,the elapsetime is about 16.7 rain with two processors, 8.8 rainwith fourprocessors, and4.9 min with eightprocesSOLS. 114th Meeting:AcousticalSocietyof America S103 THURSDAY AFTERNOON, 19 NOVEMBER 1987 UMS 5, 1:30 TO 4:05 P.M. SessionVV. Underwater AcousticsVIII: Volume and Object Scattering C. P. de Moustier, Chairman ScrippsInstitutionof Oceanography, Marine PhysicalLaboratory.4-005,La Jolla, California92095 Chairman's Introduction--l:30 ContributedPapers 1:35 Wl. Averageintensity of an acousticbeamafter propagutingthroughu turbulent ocean.J. H. Tarng and C. C. Yang (CSSL, Department of ElectricalEngineering,The PennsylvaniaState University,University Park, PA 16802) The averageintensityof an acousticbeampropagatingin a turbulent oceanisevaluatedusingthepathintegraltechnique. Because of theexistenceof the verticalsound-speed profilein the ocean,thereis more than onestationarypath for the path integralformulation.Also, thesestationary pathsarecurvedrayswhicharederivedby themethodof ray tracing. The effectsof randomfluctuations in soundspeedarestudiedby introduc- ingbothinternalwavesandtemperature finestructure.The resultsshow that the acousticbeambecomes broaderandasymmetricalongthe vertical axis.The effectsof the broadening,asymmetry,andmultiplestationary pathsare examinedin detail.Interferenceamongmultiplestationary pathsis expected.In addition,theseresultsare comparedwith thoseof othersobtainedby usingdifferentapproaches. [ Work supported by ARL, The PennsylvaniaStateUniversity.] calledp(x),withp= 1corresponding to6c= 0, wasusedasa measure of thedistancebetween p andp,. Resultsarepresented ascurvesofp (x) in dB unitsfor variousfrequencies, depending on themagnitudeofrc. It was foundthatp (x) decreases monotonically with range,on theaverage,and saturatesin rangeat a valuethat decreases with increasingfrequencyat therateof about3 dB/oct. The rangeto saturation,calledthe"predictability horizon," is surprisinglyshort for most frequencies of interestfor moderate •5c. 2:20 VV4. Modelingof acoustictransmission in the Straitsof Florida. Charles L. Monjo and H. A. DeFerrari (Departmentof AppliedMarine Physics, RSMAS,Universityof Miami,Miami, FL 33149) Acoustic transmission experiments, overa 24-kmrangein theStraits ofFlorida,havebeenmodeled usingraytheory.Threeacoustic moorings andthreethermister moorings wereoperated fora 30-dayperiod.Pulse responses (460-Hzcarrierwith100-Hzbandwidth) wererecorded every 12rain,andtemperature at severaldepthseveryhalf hour.Pulseswere 1:50 VV2. Acousticpulsepropagation in a turbulentocean.J. H. Tarng and C. C. Yang (CSSL, Department of Electrical Engineering, The Pennsylvania StateUniversity,UniversityPark, PA 16802) Temporalbehaviorof acoustic pulsepropagation in a turbulentocean isinvestigated by usingthemethodof temporalmoments.In thismethod, the zeroth, first, and secondtemporalmomentsare related to the total energyof the pulse,arrival time, and pulsewidth, respectively.To computethesemoments,insteadof the two-frequencymutualcoherence function, only the coefficients in a powerseriesexpansion of thisfunctionare required.In particular,only thecoefficients up to the second-order terms areneeded.The pathintegraltechnique,incorporated with themethodof ray tracing,is usedfor evaluatingthosecoefficients. The verticalsoundspeedprofileis takeninto account.The turbulencesin the oceaninclude the internalwavesand temperaturefinestructure.For mathematicalsimplicity,a bilinearsound-speed profileis considered, and it is assumedthat the propagationrangealonga horizontalaxisis short,suchthat the rays alwaysstayeither aboveor belowthe depthof minimumsoundspeed. More generalresultscan be extendedeasily. [Researchsupportedby ONR.] 2:05 VV3. Sensitivity of acoustic fields to environmental uncertainties. L. B. Dozier, H. Freese, E. S. Holmes, and F. D. Tappert (Science foundto be highlyvariablein traveltime,duration,andshape.The MEDUSAraytracingeigenray modelwasusedtoexplainthepulseformationandvariation,usingthethermister timeseriesasinput.Model predictions of pulseresponses agreewiththedata.Positive temperature anomalies bringaboutlongdurationpulses, 200ms,whilenegative temperatureanomalies bringaboutshortdurationpulses, 50ms.Themechanismbehindthe pulsedurationvariationwasfoundto be a functionof bottomloss,nota focusing effectof thesound-speed profile.Negative temperature anomalies tendtodirectgrazingeigenrays moresteeply into thebottom,causing greater bottomlosstothefrontendofthepulse.The front endof the 200-mspulseis "lost" belowthe noiseleveland a short duration50-mspulseresults.[Worksupported by ONR.] 2:35 VVS. Acousticmonitoringand plumemappingof drilling fluid discharge. John J. Tsai and John R. Proni (NOAA/AOML, Ocean Acoustic Division, Miami, FL 33149) An approachto monitordrillingfluiddischarge from anoil rig andto map its plumedistributionby high-frequency acousticswas proposed. Experimentswereconductedat two nearbyrigsof the East Flower GardenBank in the Gulf of Mexico,and data wereusedto providevertical plumestructureand two-dimensional distributionmapsat fixeddepth. The ability to monitor the discharged plume continuously in time and spacemakesthe acoustic technique moreconvenient andcosteffective than other conventional methods. ApplicationsInternationalCorp., 1710GoodridgeDrive, McLean, VA 22102) 2:50 APE model has been usedto compare2D cw acousticfieldsp(x,z) propagatingin sound-speed fieldsc(x,z) to a referenceacousticfield p, (x,z) propagatingin the referencesound-speed fieldcr(x,z). Samples of the range- and depth-depeodentsound-speeddifference fields, 6c(x,z) = c(x,z) -- c, (x,z), weregeneratedfrom physicalmodelsof ( 1) instrumental errors, (2) internal waves, and (3) baroclinic waves (me- soscale).The squaremodulusofthenormalizedinnerproductoverdepth, $104 J. Acoust. Soc.Am.Suppl.1,Vol.82, Fall1987 VV6. The influence of bubbles on sea surface backscatter measurements. PaulD. Koenigs,JosephM. Monti (Naval UnderwaterSystems Center, New London, CT 06320), Bernd Nuetzel, and Heinz Herwig (Forshungsanstaltder Bundeswehrfuer Wasserschall-undGeophysik, 23 Kiel 14, West Germany) The resultsfrom a recentseasurfaceacousticscatteringexperiment, 114thMeeting: Acoustical Societyof America $104 whichwasconducted in theNorthSea,arepresented withaccompanying seasurfaceroughness parameters andsubsurface bubbleinformation. The acousticdata wereobtainedutilizinga high-resolution (narrowbeamwidth) pulsedparametric sonartransmitter andconventional receiver. Scattering strengthvalueswereobtainedasfunctions of frequency(3-18 kHz} for windspeeds from2-45 knots.It appearsthat thebackscattering strengthat30øgrazingangleiscaused bythehigh-frequency wavenumber spectrumat low wind speedsand by subsurface babblesat high wind speeds. The backscattering strengthshowsstrongfluctuations in the intermediateregioncausedby both scatteringmechanisms. 3:05 VV7. Nearfield calculationsfrom elongatedobjectsand a comparison with the farfield form function.M. F. Werby, Guy Norton (NORDA Numerical Modeling Div. NSTL, MS 39529), and G. Gaonaurd (NSWS, White Oak, SilverSpring,MD 20910) It isusualto examinefarfieldformfunctions whenexamining scatteringfromsubmerged objects. In thisstudy,whathappeas whenoneexaminesthe bistaticangulardistributionsin the very nearfieldfor elongated spheroids wasdetermined. The distancefromthefieldisthenprogressively movedfrom the objectin multiplesof the semimajoraxis until the resultsagreewith the farfieldform function.It isfoundthat thisdoesnot occuruntil oneis a distancegreaterthan 20 timesthe largestdiameter fromtheobject.Whenoneiscloseto theobject,thecalculations reflectthe fact that a shadowis caston the observerand, therefore,lhe patternis ratherbroadcloseupandit narrowsasoneprogressively movesfrom the object.This lasteffectcanbeobtainedusinga simplegeometrical expressionthat agreeswith the theoreticalcalculations. incidence high-order components, that,in turn,rendertheresulting matfix problemill-conditioned. Themethodusedin thisstudyispresented with severalrepresentative numericalexamples includingobjectsof aspectratiosof 30 to I for kL/2 valuesto 120. 3'.35 VV9. Scattering from an ellipsoid submergedin the ocean. Michael D. Collins (Department of Engineering Sciences and Applied Mathematics, Northwestern University, Evanston, IL 60201) and Michael F. Werby (Naval OceanResearchand DevelopmentActivity, NSTL Station, MS 39529) A timeperiodicsourceandan ellipsoidaresubmerged in a stratified oceanseveralkilometersapart.The methodof matchedasymptotics is usedto derivean approximate expression for thefieldscattered fromthe ellipsoid.The matchedasymptotics solutionconsists of an innersolution valid near the scattererand an outer solutionvalid away from the scat- terer.The two partsof the solutionare matchedin the regionin which they are both valid. The resultingsolutionis composedof solutionsto problemsexistingmodelscan handle.The asymptoticlimit appliedto arriveat theresultrequiresthatthescatterer issmallrelativeto thedepth of theoceanandisawayfromtheoceansurfaceandbottom.However,the methodshouldbe sufficiently robustto allow theseassumptions to be relaxedgreatly.Thismethodisonlyweaklydependent onfrequency; the frequency mustbeon theorderof 10Hz or greater.[Worksupported by ONR and NORDA. ] 3:50 3:20 VV10. VVS. Eigenvalue methods in the solution of acoustical scattering problemsfrom submergedobjects.Guy Norton, Richard Keiffer, M. F. Werby (NORDA Numerical ModelingDiv. NSTL, MS 39529), and L.H. Green (Martin Marietta Baltimore, Baltimore, MD 21220) Waterman'sextendedboundarycondition(EBC) methodhasproven to beextremelysuccessful in dealingwith scatteringfromsubmerged targets.An alternateand usefulcomputational method(which is relatedto Enskog'smethod)separatelyconsiders the Heimholtzintegralequation at exteriorand interiorpoints,as doesthe EBC method.However,the interiorproblemcan be shownto transformto an eigenvalue problem, whichproduces eigenstates thatspanthespaceof thedisplacement onthe objectsurface.The eigenstates canbe useddirectlyto solvethe exterior problemandyielda numericallys•ablcandconvergent solution.This new methodavoidsproblemsencountered by the EBC method.For example, in theusualEBC approach,the unknownsurfacetermsareexpandedona known,but to someextentarbitrary,basisset,with unknownexpansion coefficients. The numberof expansiontermson the surfaceand thoseof the incidentpartialwavesmustmatch.Computationally, this requirement can often dictatethere be many more incidentpartial wavesthan strictlyrequiredfor convergence of theincidentfield.This leadsto small- S105 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 Limitations of meclmnieal struetares as wave vector filters. Y.F. HwangandG. Maidanlk(David Taylor NavalShipR&D Center, Bethesda,MD 20084) Althoughthe high-wavenumber contentin a highlysubsonic (low Mach number) turbulentboundarylayer (TBL) is knownwith reasona- blecertainty,theIow-wavenumber contentis not.In a wavevectorfilter chosen anddesigned tomeasure theiow-wavenumber contentofTBL, the majorsensitivity regionis appropriately placedin that spectraldomain. Inevitably,someminorsensitivity regionslie in the high-(convective) wavenumber region.Theseminorsensitivity regionsare thenfedby the spectralpressure ridgein theTBL. Thisfeedingmaysubstantially contaminatethemeasurements. The properchoiceandthenoveldesignof a filterisjudged,byandlarge,byitsabilityto subdue suchcontamination. In thispaperwavevectorfilters,for measurements of theIow-wavenumbetcontentof TBL, asprovidedby idealrectangular andcircularpanels, are comparedand contrasted. In thisexamination, the contamination caused bythehigh-wavenumber contentin theminorsensitivity regionsis estimatedby simulatingit as a 1ow-wavenumber contentin the major sensitivity region.Theseestimated equivalent 1ow-wavenumber contents constitutethe lower boundsthat the wave vector filters are capableof measuring. 114thMeeting:AcousticalSocietyof America S105 FRIDAY MORNING, 20 NOVEMBER 1987 TUTTLE CENTER ROOM, 8:25 TO 11:30A.M. SessionWW. PhysicalAcousticsVIII: AcousticLevitation Anthony A. Atchley, Chairman Department of Physics, Code61.4E NavalPostgraduate School, Monterey, California93943 Chairman's Introduction---8:2$ ContributedPapers 8:30 WW1. Acousticlevitation at high temperaturesin the microgravityof space.CharlesA. Roy,DennisR. Mcrklcy,GregoryR. Hammarlund, and ThomasJ. Danley (Intersonics,Incorporated,3453 Commercial Avenue, Northbrook, IL 60062) ysistakesinto accountacousticradiation,surfacetension,gravity,and compressibility effects. Salientfeaturesof thistheoryapplicable to acoustic levitationwill be presented. Calculatedshapeproperties will be comparedto existingresultsof one-dimensional theories.Also,the acoustic forcedependence onks, obtainedfromequilibriumpositioncalculations, will becompared to Hasegawa's results.[ Worksupported by NASA.] Acousticlevitationof a smallspecimen is obtainedin theenergywell produced bya singleaxisarrangement consisting ofa soundsource anda small acousticreflector.At high temperatures,the acousticforcesare 9:15 generallyinsufficient to levitatea specimen exceptin the microgravity environmentavailablein space.A singleaxisacousticlevitator(SAAL) hasbeenbuilt to levitatea specimen insidea high-temperature furnace whereit may be heated,melted,cooled,and solidifiedwhilebeingpositionedwithoutphysicalcontact.Theacoustic fieldconfiguration in sucha containerless processing devicehasbeenanalyzedandtheexpected levitationor positioning forcesarecalculated forthespecific caseoftheNASASAAL experimental hardwareasflownon theSpaceShuttleon theSTS6 IA mission.Thisexperiment successfully levitatedandprocessed three samples at temperatures from600øCto 1550øC.Experimental dataare presented andtheresults compared withthosepredicted. [Worksupported by NASA. ] 8:45 WW4. Boltzmann-Ehrenfest adiabatic principle applied to acoustic forces in a single-modelevitator. S. Putterman, JosephRudnick (PhysicsDepartment,Universityof California,LosAngeles,CA 90024), and M. Barmatz (Jet PropulsionLaboratory,CaliforniaInstituteof Technolgy,Pasadena,CA 91109) An application of theBoltzmann-Ehrenfest adiabaticprincipleyields a fundamentalrelationship betweenthe acousticpotential,actingon a sample positioned ina single-mode cavity,andtheshiftintheresonance of an isolatedmode.The theoryis verygeneraland appliesto a sampleand resonantcavityof arbitraryshapeand dimension.In the caseof a small samplein a lossless cavity,thestrictproportionality between thepotential and frequencyshift followsfrom a fairly simpleargument.One useful application of thisrelationship is thedetermination of positioning forces and torqueson a leviratedsamplefrom frequencyshift measurements. The resultsof experimentalmeasurements of the equilibriumpositionand WW2. Acousticlevitationin the presenceof gravity. M. Barmatz (Jet Propulsion Laboratory,CaliforniaInstituteof Technology, Pasadena, CA 91109) and P. Coilas (Department of Physicsand Astronomy, CaliforniaStateUniversity,Northridge,Northridge,CA 91330) orientationof a levitatedsamplewill bepresented, andtheirconsistency with theoreticalpredictionswill be discusssed. [Work supportedby NASA. ] A theoryof acousticlevitationin the presence of gravityhasbeen developed for arbitrarilyorientedchambers of rectangular, cylindrical, andspherical geometries. Simultaneous modeexcitation wasalsoincludedin theanalysis. Examples ofthetotalpotentialandforceprofilewill be givenforcommonly usedlevitation modes asa functionofchamber orien- WWS. Theory of oscillational instabilities in acoustic levitators. tation. Additional levitationpositionsare obtainedwithin the chamber CaliforniaInstituteof Technology,Pasadena,CA 91109) for appropriately chosenmodesand orientations. Levitationpositions withina chamberwheretheacousticforcesareisotropicwill bepresented for the caseof small kR. The effectsof modedegeneracyon the total potentialandforcefor thecaseof thetripleaxislevitatorflownon the SpaceShuttlewill alsobediscussed. [ Worksupported by NASA.] 9:30 JosephRudnick(PhysicsDepartment,Universityof California,Los Angeles,CA 90024) and M. Barmatz(Jet Propulsion Laboratory, A generaltheorydescribes theoscillationalinstabilities of an acousti- callylevitated sample in a resonant cavity.Thetheory,based ona Green's functionmethod,predictsthe onsetof instabilities and their saturation. Important factors controlling theseinstabilities inculde drivefrequency, spatialstructure of thelevitation mode,andthesizeof thesample. The instability in a zerogravityenvironment hasfundamentally different onsetcharacteristics thanin finitegravity.Calculations revealhysteresis and saturation effects. Manyof thepresent results areconsistent witha pre9:00 WW3. Shapeof acousticallylevitated samplesin a resonantrectangular chamber.H. W. Jackson,M. Barmatz,and C. ShipIcy(Jet Propulsion Laboratory,California Instituteof Technology,Pasadena,CA 91109) A generalizationof Laplace'sformulaapplicableto vibratingsurfaces wasusedto developexpressions for the equilibriumshapeand positionof dropsandbubblesin a resonantrectangularchamber.Expandingvarious physicalquantitiesin termsof sphericalharmonicsand applyingthe quantum-mechanical angular-momentumgradient formula leads to tractableexpressions valid for largeas well as small ks. New resultsfor volumerenormalizationof dropsandbubbleswerealsoderived.The anal- S106 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 viouslydeveloped empiricalmodelof theinstabilities [S. L. Garrettand M. Barmatz, J.Acoust. Soc.Am.Suppl.177,S21(1985)]. Typicalexamplesof theaboveeffects will bediscussed. [ Worksupported by NASA.] 9:45 WW6. Eigenfrequenclesof axisymmetric cavities: Numerical calculations.Arthur E. Woodling and James B. Mehl (Physics Department,Universityof Delaware,Newark,DE 19716) The eigenfrequencies of axisymmetric, hard-walledacousticcavity resonatorshavebeencalculatedby usingan integral-equation represents- 114thMeeting:Acoustical Societyof America S106 tionof theHelmholtzequation withNeumannboundary conditions. For axisymmetric shapes, theintegralequation is one-dimensional. Numeri- calresulnas havebeenobtained fora perfect sphere, andforslightly deformedspheres, andfor cylinders withan internalsphere onaxis.The results forslightly deformed spheres arein close agreement withthepredictions ofboundary-shape perturbation theory, fromwhichtheytypically differbylessthan6 partsin 106.Theresults forcylinders withinternal ference fieldthathascapabilities forthelevitationandpositioning ofsmall objectsor liquiddrops.The theoreticalpredictions regardingthistechniqueareexaminedandthoseresultsarecompared withlaboratorymea- surements. The natureof the apparatus andits capabilities is alsoaddressed.[Work supported by NASA.] spheres areinexcellent agreement withmeasurements [M. Barmatzetal., J. Acoust.Soc.Am. 73, 725-732 (1983)]. 10:45 10:00 WW7.Perturbation calculations ofeigenfrequencies ofcylindrical cavity with internal sphericalobstacle.JamesB. Mehl and Robert N. Hill (Physics Department, University of Delaware, Newark,DE 19716 ) Theeigenfrequencies ofa hard-walled acoustic resonator arefoundby solvingtheHcimholtzequationwithNeumannboundary conditions. A cavityCwithaninternal hardobstacle Bofvanishing sizewasconsidered. As thesizeof theobstacle shrinksto zero,theeigcnfunction doesnot approach ancigenfunction oftheemptycavity. Thusasingular perturbationtheory isrequired. A formalism wasdeveloped andtested bycalculatingtheeigenfrequencies ofa cylinder withaninternal sphere ontheaxis. Theresults arecompared withtheoretical values determined bya bound- sty-integral-equation method (A. E.Woodling, Ph.D.thesis, University WWI0. Specimentranslationalcontrolcapabilitiesusingan oppo•l radiatoracouaticlevitationsystem. DennisR. Merkley,CharlesA. Rey, Gregory R. Hammarlund, and Thomas J. DanIcy (Intersorties, Incorporated,3453 CommercialAvenue,Northbrook,IL 60062) Recentdevelopments on an opposed sound-source single-axis acousticlevitation system arediscussed. Newandimproved specimen positioningcontrolcapabilities aremadeavailable byparametric variations ofthe electronic drivesignals. Variousmethods aredescribed in whicha specimencanbeheldin onepositionor translated alongtheaxisof thelevitationbysimpleelectronic controlsignals. A technique isalsodescribed to provideincreased specimen positioning stabilityusinga closedloopsystem. [ Work supported by NASA. ] of Delaware,1986)andexperimentally [M. Barmatzet al., I. Acoust. Soc.Am.73,725-732( 1983) ]. [Worksupported in partbyNASA.] 11:00 10:15 WW8. Eigenfrequencies of a cylindricalcavity with an internal off-axis sphericalobstacle.M. Barmatz,G. Avcni (Jet PropulsionLaboratory, California Institute of Technolgy,Pasadena,CA 91109), and James WWII. Enhancedacousticlevitationusinghorn loadedand faeusing radiators.ThomasJ. DanIcy,CharlesA. Rey, DennisR. Merkley,and GregoryR. Hammarlund (Intersorties,Incorporated,3453Commercial Avenue, Northbrook, IL 60062) 19716) An improved piezoelectricdriven, focusingacousticradiator for acousticlevitationhasbeendesigned,tested,and evaluated.This device A hard sphericalobstacleis mountedinsidea cylindricalcavityat a variableposition.The resonancefrequencies of the cavity were deter- displays a significant improvement in focusing toproduce betterspecimen stabilityandenergywell shapingaswell asincreased transducer power bandwidthandelectroacoustic conversion efficiency. Corresponding ef- B. Mehl (PhysicsDepartment,Universityof Delaware,Newark, DE minedasa functionof theposition ofthesphere forobstacle positions on andofftheaxisof thecylinder.Thedependence of theresonance frequencyonthespherepositionisimportantin analyzingthestabilityofacousticallylevitatedsamples. Theresultsarein goodagreement withtheoretical calculations basedon the singularperturbationtheoryof Mehl and Hill. Themeasurements will alsobecompared to a newtheorythatutilizesthe fectsresultingfrom thehornloadingof thisacousticradiatorareconsideredas well as the benefitsof usingthe radiatorin conjunctionwith hot wall high-temperature furnaces.In orderto aid in theevaluationof these acoustictransducerimprovements, a simpleand usefultechniquefor acousticfield mapping will also be discussed.[Work supportedby NASA. ] Boltzmann-Ehrenfest adiabaticprincipleto predictthe spatialdependeneeof tlheacousticpotentialfrom the resonance frequencies. [Work supportedby NASA. ] 11:15 10:30 WW12. Enhancements of acousticlevitationin thepresence of increased gas pressure.Gregory R. Hammarlund,DennisR. Merklcy, Charles A. Rey, and Thomas J. DanIcy (Intersonics,Incorporated,3453 WW9. A newacoustic levitationdeviceusingtheinterference soundfield generatedfrom two opposedradiators.Charles A. Rey, Thomas J. Danley, Dennis R. Merkley, and Gregory R. Hammarlund (Intersonies,Incorporated,3453 CommercialAvenue,Northbrook,IL 60062) A newtechniquefor acoustically levitatinga specimen is described. The technique deviatesfrom a previoussingle-axis acousticlevitation methodthatutilizesa single-acoustic radiatoranda passive reflector to generate theinterference produced soundfield.Thenewmethodinvolves theuseoftwoopposing focusing acoustic radiators thatproduce aninter- S107 d.Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 CommercialAvenue, Northbrook, IL 60062) Acousticlevitationfor containerless materialsprocessing hasbeen investigated for usein the 1-genvironment.Enhancements havebeen obtainedby increasing thegaspressure of theacousticlevitationsystem. Interdependence of gaspressure, soundpressurelevel,acousticpositioningforce,andgastemperaturehasbeenmeasured. Also,nonlinearacoustic effectsand harmonic content within the acoustic field are examined as a functionof pressure. Finally,resultsof l-g acoustic levitationexperimentsutilizingincreased gaspressures arepresented. [ Work supported by NASA. ] 114thMeeting:Acoustical Societyof America S107 FRIDAY MORNING, 20 NOVEMBER 1987 UNIVERSITY LECTURE HALL, 8:30TO 11:30A.M. SessionXX. Psychological andPhysiologicalAcousticsVIII: BinauralHearing Leslie R. Bernstein, Chairman Department of Psychology, Psychoacoustics Laboratory, University of Florida,Gainesville, Florida32611 Contributed Papers 8:30 XX1. Ear advantagesfor monaural periodicity detection. James A. Bashford,Jr. and Richard M. Warren (Department of Psychology, Universityof Wisconsin-Milwaukee,Milwaukee, WI 53201) Monauralasymmetries werefoundfor periodicitydetectionin experimentsusingrepeated200-mssegments of Gaussiannoise(repetitionfrequency5 Hz). In experiment1,anoverallleftearadvantage wasfoundfor repeatednoisedeliveredmonaurallyandopposed bycontralateralsilence. In experiment2, lateralizationof the monauralsignalwas abolishedby simultaneouspresentationof on-line noiseto the oppositeear (contralateral inductioncausedthe monauralsignalto be perceivedascenteredon the medialplane). Althoughthis manipulationeliminatedthe possible influenceof attentionalbiasesfavoringone of the sides,ear advantages werestill obtained.Alternativemechanisms will be discussed, including possible asymmetries in activesubcortical processing of periodicityinformation. [Work supportedby AFOSR.] 8:45 XX2. Binaural envelopecorrelationdetectionas a functionof frequency separation. Virginia M. Richards (PsychoacousticsLaboratory, Department ofPsychology, University ofFlorida,Gainesville, FL 32611) Recentcomodulationmaskingrelease(CMR) experimentssuggest that listenersareableto determinewhetherthe envelopes of dichotically presented bandsof noiseare temporallysimilar(no signal)or different (signaladdedtooneofthetwobands, ofnoise).In anefforttodetermine thesensitivity to changes in envelope synchrony asa functionof thefrequencyseparation of the twobandsof noise,listeners discriminated be- masker asafunction ofthesignal onset'time. Themasker ispreceded by quietor an Nrrforwardfringeandfollowedbyquiet,an No, or Nrcbackwardfringe.Overall,theresultsarecompatible witha forwardmasking interpretation. Further,whenthesignalonsetis simultaneous with the onsetof the 20-msmasker,the Nrr forwardfringereducesdetectability. [Work supported by NSF andAFOSR.] 9:15 XX4. Further research on free-field masking. R. Wayne Gatehouse (Department of Psychology,University of Guelph, Guelph, Ontario N1G 2W1, Canada) Gatehouse(Proc. 12th Int. Congr. Acoust.,B2-7, 1986) compared free-field masked and unmasked thresholds (two-alternative forced- choiceprocedure)with a WN musketfixedat 0øwhile5 tonalprobeswere deliveredfrom 11 frontal planepositions.Maskingwasnot necessarily greatestfor masker-signal coincidence butvariedwith frequencyandsignal position.For 1.0- and 2.5- kHz signals,detectabilitywas best and approximately equalat 0øand + 90ø,whilefor theothers(0.25, 0.5, and 4.0 kHz) therewcrcgenerallylowerthresholdsassignalandmaskerseparated.The taskwasrepeatednowwith 0.5-, 1.0-,and2.0-kHz probesand themusketsetat 0øand + 45ø.Again,free-field maskingvariedwithboth signalpositionand frequency,and bestand leastdetectionwasnot solely determined byazimuthalseparation of themaskerandsignal.On average, maskingleveldifferences wereapproximately 24.0dB for 0.5-, 11.5dB for 1.0-,and 13.5dB for 2.0-kHz signals.The apparentdifferences between earphone-derived and free-fieldmaskedthresholds,which seemmore subjectto frequencyand positionalvariations,will bc discussed. tweennoisebandswhoseenvelopes wereeitheridenticalor statistically independent. The noisebandswere100Hz wide.Thebandpresented to the left ear wascenteredat 2625 Hz; the bandpresentedto the fight ear wascenteredat valuesbetween2625 and 5250Hz. Althoughtherewere 9:30 considerable individual differences, discriminations tended to be best whenbothnoisebandswerecenteredat 2625Hz. Increasingthe frequency separation betweenthetwo bandsof noiseledto poorerperformance, but the percentage of correctresponses remainedabove60. Subsequent experiments indicatethatthediscrimination isbasedonenvelope similarity ratherthanthesimilarityofthepowerspectrum. [Research supported XXS. Resolutionof pairsof simpleand complexsoundsin simulated auditory space.Pierre L. Divenyi and SusanK. Oliver (Speechand HearingResearch, V.A. MedicalCenter,Martinez,CA 94553) by NIH. ] obtaining,in 5-degazimuthalsteps,externalear transferfunctions for bothearsofan artificialheadplacedin ananechoic room.Whenanarbitrarysoundispassed throughthedigitalfilterdefined bytheleftandright transferfunctions corresponding toa givenangle,thesoundwillacquirea 9:00 XX3. Effects of forward masker fringe on binaural detection. B. D. Simpson and R. H. Gilkcy (SignalDetectionLaboratory,Central Institute for the Deaf, Saint Louis, MO 63110) Yost [J. Acoust.Soc.Am. 78, 901-907 (1985) ] foundthat detectability of a 20-msdichoticsignal(S•r) in a 20-msdioticnoise(No) wasnot affectedby thepresence of a 500-msdichoticforwardfringe(Nrr). However, Kollmeier and Gilkey [J. Acoust. Soc. Am. Suppl. I 74, S39 (1982)] variedtheonsettimeof a 20-msS•' signalin a 750-msnoisethat switched,after 375 ms, from N•r to No, and found that the N•r noise reduceddetectabilityevenwhenthe signalwaswell into the No noise. Theyconcluded that the N•r noiseactedasa forwardmasker.The present studyinvestigates detectability ofa 500-Hz,20-msS•rsignalin a 20-msNo $108 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 A simulated acoustic replicaof a soundsourcemovingalonga circle with a 4-m radiusin the frontal horizontalhalf-planewasgeneratedby subjective azimuthcomparableto whichthe transferfunctionwasmeasured[J. BlauertandP. Laws,Acustica29, 273-277 (1973) ]. This tech- niquewasusedto measure resolution thresholds for pairsof amplitudemodulated aswellaspuresinusolds. With thelistenerrequiredto identify the relative location of the two soundsin a two-alternative forced-choice paradigm,in onesetof experiments the two soundswerefixedandthe minimumaudibleangularseparation wasobtained,whereas,in another set,the angularseparation wasfixedand the minimummodulation-or carrier-frequency separation wasmeasured. Generally, bothkindsof resolutionare better around0 azimuth than at the sides:In the "auditory fovea,"simultaneous resolution for puretonesandAM sounds(a 3- or 6kHz cartiersinusoidally modulatedinthe 150-to 400-Hzrange)isalmost asgoodasthejndobtained intheusualsequential presentation. Thepoor- estspatialresolution wasobtained fora pairof6-kHzcarriers modulated 114thMeeting: Acoustical Society ofAmerica S108 by a 100-anda 400-Hz sinusoldandthe bestfor pairsof AM toneswith differentcarriers.[Work supportedby the VeteransAdministrationand by a NATO TravelFellowship. ] 10:00 XX7. Psychophysical factorsof spatial impression.Frans A. Bilsen, Herman W. Kruysse, and JohanRaatgever (Applied Physics Department, Delft University of Technology, Delft,The Netherlands) Spatialimpression (or spaciousness) is animportant perceptual attributein roomacoustics, implying( 1) the sensation of a listenerin a concert hallof beingenveloped bythesoundfieldand(2) theperceived broadening of the soundsource.A psychophysical experiment will be 9:45 reportedusingdichoticpresentation of (filtered) white noise,in which XX6. The enhancement of binauralsourcelocationby scubadivers. Stewart A. L. Glegg, JohnKloske, ChrisHansen,and Terry R. Johnson(Departmentof Ocean Engineering,Florida Atlantic University,BocaRaton, FL 33431) This paperwill describethedevelopment of a passivesourcelocalization device to enhancethe ability of scubadivers to locate acoustic sources. The deviceconsists of a hat that is wornby thescubadiversand providesa distinctnull in the observedacousticfield whenthe diver is facingtheacoustic source. Testswerecarriedoutin a 25-ft-deep,198-ftdiam tank at the EPCOT Centerin Orlando,FL on 25 subjects. The resultsof thesetestsshowthat thereis a clearimprovement usingthis device,whichiscomfortable to wearandsuitablefor mostapplications. theinfluence on spatialimpression wasinvestigated of threesignalvariables,viz., theinterauralcorrelation,thecenterfrequency, andtherelativebandwidthof thefilteredversions of thenoise.Eightuntrainedsubjectsparticipatedin a scalingexperiment: They werepresented 4 times with 80 (4.5.4) experimentalconditions;subjectivecriterionwas the perceived widthof thesoundimage.Therawordinaldatawereanalyzed by canoniccorrelationanalysis(CANALS); on the resultingdata, ANOVA wasapplied.The mainoutcomeof the experimentis that the interauralcorrelation andthecenterfrequency areindependent andadequatepredictorsof spatialimpression. This resultandtheresultsof other experiments will beinterpreted in thecontextof existing binauraltheories.[Work supported by theNetherlands Organization for theAdvancement of Pure Research(ZWO). ] 10:15-10:30 Break 10:30 XXS, Effectsof rovinglevel onbinauraldetectionand discriminationon andoff l•Mdline. J. KoehnkeandH. S. Colburn(Research Laboratory of Electronics, Massachusetts Institute ofTechnology, Cambridge, MA 02139 and BostonUniversity,48 CummingtonStreet,Boston,MA 02215) falsealarm rates,whereassignificantcorrelationsfor both rateswere re- portedfor thewidebandexperiment. The lackof effectof targetphaseon hit rate in the narrow-band NoS•' condition is inconsistent with models basedon lateralpositionbut consistent with the EC modelandwith models usingthe sumof the squaresof interauraltime and intensitydifferences.[Work supportedby NIH.] NoS•rdetection thresholds andinteraural timeandintensity jnd's weremeasured bothwithandwithouta 10-dBrandomroveappliedtothe 11:00 overalllevel.The stimulusfor discriminationandthe maskerfor detection werea l/3-octnoise bandcentered at500Hz;fordetection, thetargetwas a 500-Hztoneburst.Thresholds andjnd'sweremeasured using anadaptive,four-interval, two-alternative forced-choice procedure forseveral interaural timeandintensity reference conditions including differences of0 and24dBand0 and600gs.Results indicate noeffectofrovinglevelfor the0-dB,0-psreference conditionfor binauraldetectionor interaural discrimination. ForNoS•'detection, thereisnoeffect of rovinglevelor reference conditionon thresholds. For interauraltimediscrimination, jnd'sarelargeroffmidlineandrovinglevelvariationhasnoconsistent effect onperformance. Forinteraural intensity discrimination, jnd'sgenerally increaseoff midlineand are usuallylargerwhenmeasuredwith rovinglevelthanwithout.[Work supported by NIH.] 10:45 XX9. Effectsof target phasein narrow-bandfrozennoisedetectiondata, S.K. Isabelle andH. S. Colburn(Biomedical Engineering Department, BostonUniversity,110 CummingtonStreet,Boston,MA 02215 and Research Laboratory of Electronics, MassachusettsInstitute of Technology, Cambridge,MA 02139) Hit and falsealarm rateswere estimatedfor detectionofa 500-Hz tone XX10. Lateralization on the basisof interauralenvelope delays:The effectof additional frequency components. RaymondH. Dye,Jr. and AndrewJ. Niemiec(ParrelyHearingInstitute,LoyolaUniversity, 6525 North SheridanRoad,Chicago,IL 60626) A lateralization experiment wasperformed in whichthresholdinter- auralenvelope delays weremeasured asa function ofmodulation frequency (fm = 25, 50, 100, 200, 300, 400, and 500 Hz) for three- and five- component complexes whosecenterfrequency• } waseither2000 or 4000 Hz. A two-alternative forced-choice procedure wasusedin which the envelopelaggedto the right ear duringoneintervaland to the left intervalduringtheother.The levelof eachcomponent was50 dB SPL, andthesignaldurationwas200mswith 10-mslinearrise-decay times. Thresholds obtained with three-andfive-component complexes were quitesimilarregardless offro, a findingthatseemed consistent withthe notionthattheenvelope isextracted fromcomponents interacting within anauditory filter,withmoredistalcomponents having noeffect. Surprisingly,makingtheoutermost sidebands (f• - 2f• andfc + 2f• ) riotic wasfoundtoseverely impareone'sabilitytoutilizeenvelope delays even whenthef• 's werequitehigh.Thesefindings placein doubtthecontentionthatthebinaural auditory system extracts envelopes bymonitoring the outputsof narrow-band auditorychannels.[Work supported by NINCDS and AFOSR. ] targetin eachoften samples of l/3-oct noisecenteredat 500Hz for both NoS•randNoSoconditions. Theeffects oftargetphase(0øand90ø)onhit ratewereinvestigated forthreesubjects. Resultsforthisnarrow-band case weresimilarto previously reportedresultsforwidebandmaskers[Gilkey etal., J. Acoust.Soc.Am. 78, 1207-1219(1985) ]. Two differences were noted.First,forNoS•'conditions, theeffectofsignal phase wasnotsignificantforthenarrow-band maskers, whereas a smalleffectwasreported for wideband maskers. Second,no significant correlation wasobserved between NoSo and NoS•r results across masker waveforms for hit and S109 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 11:15 XXI1. Interaural envelope correlation and the high-frequency MLD. Leslie R. Bernstein (Psychoacoustics Laboratory, Department of Psychology,Universityof Florida, Gainesville,FL 32611) Data from a numberof investigations suggestthat listenersare extremelysensitiveto interauraltemporaldisparitiesin the envelopes of 114thMeeting: Acoustical Society ofAmerica S109 configurations includingNoSo,NoSre,andNoS•rat a centerfrequency of 3571Hz. In addition,a sinusoidally amplitude-modulated 3571-Hz tone dioticmaskingnoiserangedfrom50to400Hz. For all conditions, dichotic thresholdswere accountedfor quite well by assumingthat, for each listener,detectionoccurswhenthesignaldecreases theinterauralcorrelationof theenvelope of thenoiseto somethresholdvalue.This findingas wellasalternative explanations will bediscussed in termsofmodernmodelsand theoriesof binauralhearing.[Researchsupportedby Air Force whosecarrier wasinteraurallyin phasebut whosemodulatorwas inter- Officeof ScientificResearch.] high-frequency, complex waveforms. In thepresent study,anattemptwas madeto assess whetherthis sensitivitycouldbe the basisfor MLDs obtainedat high frequencies. An adaptive,two-alternativeforced-choice task was used to estimate detection thresholds for a number of interaural aurallyphase-reversed wasemployedas a signal.The bandwidthof the FRIDAY MORNING, UMS 4 AND 5, 8:30 A.M. TO 12:00 NOON 20 NOVEMBER 1987 SessionYY. UnderwaterAcousticsIX: Bottom Interacting Ocean AcousticsI David R. Palmer, Chairman NOAA--AtlanticOceanographic andMeteorological Laboratory, 4301Rickenbacker Causeway, Miami,Florida33149 Chairman's Introduction--8:30 Invited Papers 8:35 YYI. Acousticiasuesfrom a marinesurveyor'sviewpoint.WesleyV. Hull (Chartingand GeodeticServices, National Ocean Service,NOAA, Rockville, MD 20852) Theships andinstrumentation usedin theNationalOceanService hydrographic andbathymetric surveyingrepresent capitalinvestment of severalhundredmilliondollarsandrecurringannualcostsof tensof millions.Theefficiency in surveycoverage, whilemaintaining requiredsurveyaccuracy, isof utmostconcern. Deficiencies inanyoftheseveral typesofacoustic systems usedinthesurveys cananddocontribute significantly tothelossofsurvey production. Thispaperdescribes brieflythreetypesofsurvey missions, highlighting the acoustic systems usedandtheknownor preceived acoustic phenomenological factorsthatcontribute to lost production. Thesesystems include dualfrequency echosounders, multibeam swathsonarsystems, andconventionalsidescan.The paperalsodiscusses emerging technology, now underconsideration, includingboth intefferomctricand multibeamsidescansystems,againwith emphasison the critical acousticphenomena. Finally,a newconcept dubbed "theacoustic knife,"forwhichanexperimental project isnowcontemplated, is discussed. 9:00 YY2. Scattering of soundfromtheroughseafloor. T. K. Stanton(Departmentof GeOlogy andGeophysics, Universityof Wisconsin,Madison,WI 53706) Thescattering of soundfromtheroughseafloor andresultantechofluctuations arefunctions of seafloor geomorphology andsonarparameters suchasbeamwidth, frequency, andangleofincidence. Thescattering is verydifficultto describe analyticallyandquiteoftenrequiresnumericalandempiricaltechniques. Like most roughinterfaces, theseafloor canbedescribed bytwoclasses ofsurfaces: ( 1) stochastic and(2) deterministic. For bothtypesof surfaces, it is essential to modeltheroughness asbeingtwo-dimensional, i.e.,notstriated. Because of therandomness of theseafloor,theechowill fluctuatefrompingto pingasthesonarmovesoverthe floor.Patchesor distinctgeomorphological featureswill causedeviationsin the statisticalproperties of the echoes.[ Work supported by theONR. ] 9:25 YY$. Inversemethodsin oceanbottom acoustics.GeorgeV. Frisk (Department of OceanEngineering, WoodsHole OceanographicInstitution,WoodsHole, MA 02543) A reviewispresented ofexactandperturbative inversemethods usedtodetermine theacoustic properties of thebottom.The discussion concentrates primarilyoninversions for propertiesin thetopfewhundredmetersof sedimentusinginputdatain thefrequencyrange25-500 Hz. The one-dimensional, horizontallystratifiedcase is emphasized,althoughimplicationsfor problemsof higherdimensionalityare discussed. Descriptionsof theoreticalmethodsandexperimental techniques aswellasexamples of inversions of syntheticandexperimental data are presented. Sl 10 J. Acoust. Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting: Acoustical Societyof America $110 9:50 YY4.Bottominteraction effects onbroadband signals. PaulJ. Vidmata• (Science Applications lnternarional Corporation,McLean,VA 22102) Thispaperexamines recentprogress towardunderstanding theeffectsof theseafloor on low-frequency broadband signals. First,themajorbottominteractionprocesses atfeetinglow-frequency signals arereviewed anda simplified(singlelayerfluid) geoacoustie profileisshownto accurately predicttheireffects. A ray-based approachfor simulating singlebouncebroadband signalsis thendiscussed. The simulatorusesthesimplified geoacoustic profileto includetheseaflooraspart of thepropagationmedium.Underconditionsfor whichthe simplifiedprofileaccuratelyrepresents the acousticsof the seafloor,calculatedwaveformsagreewell with acousticdata. Next, an approachfor simulatingmultibounce signalsinteractingwith a layeredseaflooris described and usedto showthat layeringin the seaflooris the likely causeof somecharacteristics of acoustic datathatarenotpredictedbythesinglelayerprofile.Finally,thetimespreading ofbroadbandsignalsin anarea with about200 m of sedimentcoverisexamined.Acoustics datahavebothsmall-scale( < 200 ms) andlargescale( > 1 s} timespreads. Viewingthedatathroughseveralfrequency bandsrevealsthat small-scale time spreads areconcentrated at highfrequencies (consistent with reflectionfor nearsurfacelayering)whilelargescaletimespreads arecomposed primarilyof lowfrequencies (consistent withinteractionwiththesubstrate). A simplerayanalysis showsthatthescattering fromtheregionnearthepointof specularreflection cannotbe thesolecause ofthelarge-scale timespreading. a•OnleavefromAppliedResearch Laboratories, TheUniversity of Texasat Austin,Austin,TX 78712. ContributedPapers 10:,15 YYS. Stress-inducedanisotropy in sediment acoustics. R. D. Stoll (Lamont-DohertyGeological Observatory, Palisades, NY 10964} A modelto describeseismicwavepropagation in bothdry andwatersaturated particulatematerialsisderivedonthebasisof theresponse ofa regulararrayof like elasticspheres.The theoriesof Hertz andMindlin are usedto modelthenormalandtangentialcompliances at theintergranular contacts,andtheoverallresponse of thearrayisfoundto begoverned by a setofnoalineardifferentialequations. Theseequations areintegratedover a particularstresshistorythat simulatesthe normalbuildupandreliefof geostaticstress(overburdenpressure ) in nature.Wavevelocitiesare evaluatedat variousstages duringthisstresshistoryandtheresponse isfound tobeanisotropic with respectto bothP andS wavevelocities. In orderto simulatethe response of marinesediments, theeffectsof watersaturation are investigated by usingthe compliances calculatedfor the dry skeletal framein conjunction with the Biot theory.Finally, the predictions of the modelarecomparedwith recentlaboratoryexperiments andfoundto be in generalagreement.[Work supportedby ONR, Code 1125OA. ] 10:30 10:45 YY7. Quantitative experimentalverificationof seabedshear modulus profile inversionsusingsurfacegravity (water) wave-inducedseabed motions. Mark Trevorrow, Tokuo Yamamoto, Mohsen Badiey, Altan Turgut,and Craig Conner(Divisionof AppliedMarine Physics, Rosensriel School of Marine and Atmospheric Science, 4600 Rickenbacker Causeway,Miami, FL 33149-1098) A previously developed theoryenables seabed shearmodulusprofiles to becalculatedthroughthecombination of waterwave-seabed interactiontheoriesandgeophysical inversion methods. A newinstrumentation system,consisting of threeorthogonally mountedaccelerometers and a pressure sensor, hasbeendeveloped tomeasure waterwave-induced pressuresandseabed motions.Thisnewpassive remotesystemprovides a fast and convenient method to determine in situ sediment shear modulus with depthprofiles,withoutdisturbing thesediment or drilling.Experiments havebeenconductedin varioussediments and waterdepthsoffthecoast of New Jersey.Extensive experimental datareductiontechniques aredeveloped to dealwithrealoceandata.Actualexperimental results aregiven,alongwithquantitative measures of uncertainty andresolution. Comparisonsbetweenthe experimentalshearmodulusinversions,in situ sediment strengthtests,andlaboratoryanalysis ofcoresamples showthat thismethodcanaccuratelypredictshearmodulusto depthsof 50 m or more.[Work supported by ONR. ] YY6. Shallow seismoacousficexperiments using shear waves. R. D. $toll, G. M. Bryan, R. Flood, D. Chayes, and P. ManIcy (Lam0nt-Doherty GeologicalObservatory,Palisades,NY 10964) Duringthe summerof 1986a seriesof seismoacoustic experiments werecarriedoutin shallowwaterofftheNewJersey shore.Thepurpose of theseexperiments wasto measurethegeoacoustic propertiesof theocean sedimentsthat comprisethe upperfew hundredmetersof the sediment column.Scismiesourcesand receiversweredeployedat or verynearthe bottom in order to excite shear waves in the sediment and minimize the effectsof interferencefrom water-bornepropagation.The experiments were performedat severalsiteswhere prior field work had established physicalproperties anda detailedprofileof thesediments. By usingconventionalair gunsdeployedin an unconventional way, stronginterface and divingshearwaveswere generated,and thesedata were invertedto obtainshearwavevelocityas a functionof depth.The inversionresults 11:00 YYg. Acousticwave propagationthroughporousmedia with arbitrary pore-sizedistributions. Tokuo YamamotoandAltan Turgut (Division of Applied Marine Physics, Rosenstiel School of Marine and AtmosphericScience,4600 RickenbackerCauseway,Miami, FL 33149} In the Biot theory, the effectof frequencyon the oseillatoryviscous forceswithin a porousmediumis treatedby replacingthe kinemarieviscosityv by an oscillatoryviscosityvF. Here, F is a functionof angular frequency•o,the kinemarieviscosity•, and the singleporesizea. In this paper,a mathematical expression of Fis presented for arbitrarydistributionof poresizesthatcanbeusedin theBiottheorywithoutmodification. It isshownthatporousmediawith a givenpermeabilityandporositymay were then compared with the predictions of a geoacoustiemodel that be representedby an infinite number of pore-sizedistributions. The dis- accountsfor the effectsof voidsratio, overburdenpressure,and other physical parameters. The in situmeasurements fromexperiments andthe gradientspredictedby the modelwerein goodagreement, suggesting a strongdependence of velocityon overburdenpressurenear the watersedimentinterface.[Work supportedby ONR, Code 1125OA]. persionand attenuationof acousticwavesthroughsuchporousmediaare independentof the pore-sizedistributionat the low- and high-frequency limits,whiletheyare stronglydependent on the pore-sizedistributionin the intermediatefrequencyrange.For porousmediawith •-normal pore sizedistributionshavinga givenvalueof permcability,the maximumspe- $111 J. Acoust.Soc. Am. Suppl. 1, Vol. 82, Fall 1987 114th Meeting:AcousticalSocietyof America S111 cificattenuationdecreases and the bandwithof dispersion increases with an increasingstandarddeviationof the pore-sizedistribution.Finally, the generalized Blot theoryfor arbitrarydistributionof poresizesis comparedwith the entiredata of compressional waveattenuationthrough surficialmarinesediments givenby E. Hamilton.Comparison betweenhis data and the predictionsfrom thistheoryshowgoodagreement.[Work supportedby ONR. ] 11:15 YY9. Effectsof seabedproperties on acousticnormalmodepropagation in a stratified shallowocean. Mohsen Badicy (Division of Applied Marine Physics,RosensticlSchoolof Marine and AtmosphericScience, thata simpleshearsourcein a layeredmediumdoesnot fullyexplainthe observed data.Syntheticsources mimicthe experimentalsourceby combininga dipoleshear,a torque,and an explosionsource(in the water). This complexsourceis appliedto differentvelocity/attenuation models, andresultsarecompared to thedata.Thefrequency bandofthesynthetics is 2-15 Hz, matchingthepredominant frequencies of thedata;thephase velocityrangeis0.05to 1.0km/s;all possible modeswithinthesefrequency andphasevelocityrangesarecomputed. Varyingthickness andvelocity of a low-velocitylayerbothat the top andwithinthe sedimentsection producestrikingchanges in thecharacterof theseismograms. Normal or inversedispersion characterize the late-arrivingguided-wave portionof the synthetics.The early-arrivingdivingwavesgenerallydo not exhibit appreciable dispersion.[ Researchsupportedby ONR. ] 4600 RickenbackerCauseway,Miami, FL 33149) Thepressure fieldof a pointsource in a shallowstratified oceanovera nonhomogeneous anisotropic seabedis calculatedusingnormalmodes. Thesediment ismodeledusingtheBiot-Yamamototheoryofporo-elastic media.The influence of geoacoustic properties of sediments on low-fre- quency transmission lossisinvestigated. It isfoundthatthevariationof permeability haslargeinfluence on acoustic waveattenuation, whereas theelasticproperties variationhaslesser effect.Finally,theexisting model is usedin an attemptto extractthe depthstructureof permeability and porosityof theseabed. [ Worksupported by ONR. ] 11:30 YYI0. Syntheticmodelingof shearwavesin shallowmarinesediments.J. A. Carter, G. H. Sutton, N. Barstow (Rondout Associates, Incorporated, P.O. Box224,StoneRidge,NY 12484),andJ. I. Ewing (WoodsHole Oceanographic Institution,WoodsHole, MA 02543) Full-waveform syntheticseismograms arebeingusedto aidinversion for shearvelocityand attenuationmeasuredin shallowmarinesediments offthecoastof southernNew Jerseyin May 1986.Horizontallyandverti- cally polarizedshearwaveswereproducedby a pair of air guns,fired alternatelyleft andright,mountedona sled.Initial studiesdemonstrated FRIDAY MORNING, 20 NOVEMBER 11:45 YYI1. Dynamic behavior of sedimentand water-sedimentinterface. Homogenizedprocessapplicationto poroussaturatedmedia.C. Avallet (Metravib R.D.S., 64, ChemindesMouilles,69130 Ecully, France) An applicationof the homogenized processto thedynamicbehaviorof marinesediments ispresented here.The mediumisconsidered asa porous viscoelastic solidsaturatedby a viscousfluid. Underthe assumption of periodicityfor pore structure(microscopicscale), the homogenization techniqueleadsto the macroscopic description.The setof equationsso obtainedissimilarto theclassical Biot'sresults.The homogenized model is somewhateasierto use:A generalizedDarcy's law governsthe motion of theporefluidrelativeto the solid[ Auriault et al., J. Acoust.Soc.Am. 77, 1641-1650 (1985)]. A reasonable assumptionallowsthe computationof thefrequency dependence of thepermeability fromits quasistatic value.Using the homogenized set of macroscopic equations,compressionaland shearwavesin porousmedia,surfacewavesat fluid-porous solid interface,and reflectionpropertiesof plane acousticwaveswere calculated.Resultsarecomparedto thoseobtainedby Biot'stheoryand to experiments. The aim of this work wasto showthe applicabilityof the modelto unconsolidated sediments(high porositysilt andclayor sandy media). [Work supportedby D.R.E.T.] TUTTLE 1987 SOUTH ROOM, 9:00 TO 11:40 A.M. SessionzZ. Noise VI: Measurement Techniquesand a Review of Noise Effects on Man Julia D. Royster, Chairman EnvironmentalNoiseConsultants, Inc., P.O. Box 144, Cary, North Carolina27511 Chairman's Introduction--9:00 ContributedPapers 9:05 ZZl. The influence of the center frequency of a narrow-band shortduration impulse noise on the growth of TTS. Chantal Laroche and Raymond H•tu (Grouped'acoustiquedel'Universit•deMontr•al, C.P. 6128, Suet. A, Montr/-al, QuebecH3C 3J7, Canada) The foilowingstudyhascontributedto theknowledgeof the effecton humanhearingof thespectralcontentof impulsenoise,a parameterthat hasbeenlittle studiedto date.In thisstudy,signalswerecomputergenerated to obtain I octaveband wide impulsesfrom a (sin x)/x numerical Sl 12 J. Acoust.Soc.Am.Suppl.1, Vol. 82, Fall 1987 function.The centerfrequencywasvariedsoasto generatefour signals centeredon 1, 2, 3, and 4 kHz. In this study, ITS growth curveswere obtainedwith a targetasymptoticeffectof 8-12 dB after 30-45 min of exposure ata rateof 1pps.Therequired levelofexposure wasdetermined to obtainthistargeteffect.To reachthesamelevelof ATS, theLOof the 1- kHz signalhadto beat least5-7 dB higherthanthatof the2-kHzsignal. For twoof foursubjects, comparable levelswererequiredfor the3- and4kHz signals toobtainthepredetermined effect,butthesewereat least5 dB belowthecorresponding levelforthe l-kHz signal.Theseresultsdemonstratethatthespectral contentofimpulse isa crucialfactorin thepredictionofthenoxiousness of thistypeofnoise.[ Worksupported by IRSST.] 114thMeeting:AcousticalSocietyof America Sl 12 combined. The procedure for automatically identifyingprominentdiscretetonesperANSI S12.10isdiscussed. ZZ2. Digital processingof serosonic signals. W. C. Meetham, E. Wildauer (Department of Mechanical,Aerospace,and Nuclear Engineering,Universityof California,Los Angeles,CA 90024), and S. A. Mclnerny (AerospaceCorporation,El Segundo,CA 90009) 9:50 ZZ4. Noise sourceidentificationon an IBM typewriter using sound Experimentaldatawereobtainedfromtestsperformedin theNASAAmes7-X 10-ftwindtunnel.A digitizingprocess for flowovera bluntedgedair foil wasusedto comparedatawithananalogsystem. The nearfieldsurface-pressure signalon the wing and the farfieldaerosonicsignal werecross-correlated. Digitalfilteringwasappliedto thedatato suppress unwantedsignals. The resultsof near-/farfieldcrosscorrelationsshow, usingthedigitizedtreatment,thatthereissubstantial ringing,a relatively narrow-band source.Analogprefiltering exhibitsa shiftto lowerfrequency, dueto frequencyroll-offof the RC analogfilters.It is shownthat the clip-correlation(neededfor analog) and the digitizedfull correlation methodsgivesimilarresults.But the absolutevalueof the correlationis lostin clipping.[Work supported by NASA-Ames.] intensitytechnique. Li L ZengandMalcolmJ. Crocker(Departmentof MechanicalEngineering, AuburnUniversity,Auburn,AL 36849) Noise source identification was carried out on an IBM ball element electrictypewriter.Theselective operationmethodwasusedtosurveythe vibrationandsoundpressure of differenttypewritersources. The sound powerradiationof differentnoisesources wasmeasured usingthe twomicrophone soundintensitytechnique. The measurements showedthat theimpulseradiationof theballelementstrikingtheplatenisthehighest noise source,and the differencebetween the results with and without paperis small.Severalindexesthat suppose to qualifythe accuracy of soundpowermeasurements usingthe soundintensitytechniquewill be discussed. [Work supportedby IBM, Austin,TX.] 10.'05 9'.35 ZZS. Determination ofloudness witha programmable real-timeanalyzer. M. Brechbuehiand Svein Arne Nordby (NorwegianElectronics- ZZ3. Measuring the soundpower of a noisesourceand detectingpure toneswith a singleinstrumentcombiningtrue 1/3 octaveband digital Scantek,Inc., 51 Monroe Street,Suite i 603, Rockville,MD 20850) filters and an FI• analyzer. Ole-Herman Bjor and Gustav Ese (NorwegianElectronies-Scantek, Inc., 51 Monroe Street,Suite 1603, An efficientprogramfor determination of loudness as specified by Zwicker[E. PaulusandE. Zwicker,"Computer programs forcalculating Rockville, MD 20850) loudnessfrom i/3 o•tave band levelsor from critical band levels" (in German), Acustica27, 253 (1972)] ismadetooperateona portablerealtimeanalyzer.The programcontrolsthemeasurement andcalculates the loudness asperISO R 532B.The usermayviewtheresultsgraphically on In thepast,therehasbeena requirement for determination ofboththe soundpowerof sourcesand an investigationof the pure-tonecontent. Thisisespecially truein thecomputerandbusiness equipmentindustry. Measurementof the soundpowernormallycallsfor standard(ANSI) classIII 1/3 octaveband filters,while pure-tonedeterminationrequires narrow-b•d (FFT) measurements; hence,two separateinstrumentsysternswererequired.This paperdescribes a newinstrumentfor thesedeter- the RTA's black and white monitor or in full color on an external monitor. Thiscompletely automaticsystem operates withouttheuseof anexternal computerdue to the uniqueinternalBASICprogrammability of the realtimeanalyzerused.The usermayeasilymodifytheprogramto meethis/ her ownrequirements incorporating EPNdB, Stevens Loudness, etc. minations in which the true standardized !/3 oct filters and the FFT are 10:20-10:25 Break 10:25 ZZ6. E 'alnationof a hearingconservationprugramat a large industrial company. WilliamW. Clark,CarlD. Bohl,LisaS. Davidson,andKeith A. Melda (Central Institute for the Deaf, 818 S. Euclid, Saint Louis,MO 63110) The hearingconservation program(HCP) of a largeindustrialenmpanywasevaluated by comparing theaudiometric dataof twogroupsof workersfromthreegeographically separate plants.Group 1 wasselected from individualsemployedin areaswhere the time-weightedaverage (TWA} soundlevelexceeded 90 dBA andgroup2 fromindividualswith TWAs of less than 85 dBA. Differences between mean STS values for groupsandplantswerethemaincriteriausedin makingjudgments onthe HCP effectiveness. in addition,age,sex,yearsof service,andnonoccupational noiseexposurewereevaluated.A significant differencein mean STSwasfoundbetweenthetwo groups;however,the"nonoccupationally exposed" grouphadmoreSTSthanexpected. The male-femaledistribution,previous hearingloss,andnonoccupational exposure werejudgedto becontributing factorsin thedifferences that existedbetweenplantsand groups.[ Worksupported by GrantOH 02128from NIOSH. ] The sensitivity of predictedgroundnoiselevelsfor advancedturbopropaimrafttochanges inatmospheric absorption losses isreviewed, and theresultingcommunitynoiseenvironments evaluatedin termsof possiblecriteriafor acceptable levels.Atmospheric absorption coefficients at highaltitudesare subjectto possible errorsof over + 70% due to the currentuncertaintyin atmospheric absorptionpredictionmodelsat the very low humiditiesinvolved.However,the integratedabsorptionloss over the entirepropagationpath is nearlyindependent of the particular predictionmodelchosen. The humidityandtemperature structureof the atmosphereis a more significantsourceof variationin the total atmosphericabsorptionloss.Basedon a referenceprofileof humiditycontent versusaltitude,andexistingmodelsfor a standardatmosphere, estimates of enroutenoiseof advancedturbopropaircraftare madeandcompared with availabledata. The noiseexposureon the groundis considen:din termsof boththe maximumsoundlevelandsoundexposure levelfor one overflight,and theday-nightaveragesoundlevelfor variousscenarios of dailyoperations. Theseestimates of the resultingcommunitynoiseexposureare comparedto possible criteriafor acceptable levelsfor singleand multipleaircraftflyovers.[This work wassupportedin part by the Federal Aviation Administration.] 10:55 10:40 ZZ7. Soundpropagationand communitynoise exposureconsiderations for enroute noise of advancedturboprop aircraft. John Weslet (Wylc Laboratories,2001 JeffersonDavis Highway, Suite 701, Arlington, VA 22202) and Louis C. Sutherland (Wylc Laboratories,128 Maryland Street,El Segundo,CA 90245) Sl 13 J.Acoust. Soc.Am.Suppl.1,Vol.82,Fall1987 ZZS. A theoretical interpretation of a dosage-effectrelationshil•for the prevalence of annoyance in a community. Sanford Fidell, David M. Green, and Theodore Schultz (BBN Laboratories Incorporated, P.O. Box 633, CanogaPark, CA 91304-0633) A dosage-effect relationship by Schultz["Synthesis of socialsurveys on noiseannoyance,"J. Acoust.Soc.Am. 64, 377•,05 (1978)], that 114thMeeting: Acoustical Society ofAmerica Sl 13 relatescommunitynoiseexposureto the prevalence of annoyance, has becomea widely usedtool in assessing environmentalnoiseeffects.BecauseSchultz's relationship isappliedtoa broadrangeofexposure conditions,andbecause it isentirelyempirical,itsapplicationto specificsituationsisfrequentlychallenged. A simplemathematical modelcanaccount notonlyfor theshapeof therelationship synthesized by Schultz,but also for the differences betweenSchultz'srelationshipand that proposedby Kryter ["Community annoyancefrom aircraft and ground vehicle noise,"J. Acoust.Soc.Am. 72, 1222-1242(1982)]. A singleparameter, representing response bias,suffices for thispurpose.[Thisworkwassponsoredby the U.S. Air ForceNoiseand SonicBoomImpactTechnology programunderContractF33615-86-C-0530. ] 10 000 Hz in stepsof 1/3 octave.Plotswereobtainedin the horizontal planefor threeconditions:( 1) earopen;(2) earoccludedwith a malleablefoam-typeear plug;and (3) earcoveredby a conventional earmuff. Polardatawereconverted to field-to-drum transferfunctionsfor every 45' changein angleof signalincidencefor eachcondition.The diffuse- fieldear-open transfer functions willbecompared topreviously published functions forKEMAR. Theeffects ofearocclusion withmuffs,caps,and plugsas seenby attenuationand canalresonanceshift will be discussed. 11:25 ZZI0. A theoretical model of the annoyanceof individual noise intrusions. Sanford Fidell, David M. Green, and Karl S. Pearsons (BBN Laboratories, Inc., P.O. Box633,CanogaPark,CA 91303) 11:10 ZZ9. Effects of hearingprotectorson headdiffraction.JohnR. Franks (Bioacoustics and OccupationalVibration Section,PhysicalAgents EffectsBranch,Division of Biomedicaland BehavioralScience,National Institutefor OccupationalSafetyand Health, 4676 ColumbiaParkway, Cincinnati, OH 45226) The diffractionby theheadandpinnaof soundincidentupontheear hasbeenwell documented.However, the modificationof headdiffraction patternsby the placement of hearingprotectors in, or over,theearshas notbeeninvestigated. Polarsensitivityplotswereobtainedin an anechoic soundfieldfromtherightearof theKnowlesElectronics Mannequinfor Acoustic Research(KEMAR). Test frequenciesranged from 100- FRIDAY MORNING, 20 NOVEMBER 1987 Self-reports of noise-induced annoyance canbe modeledastheproductofa rational,decisionlike process in whichpeople discriminate among annoyingand nonannoying noiseintrusionsheardin the presence of a reference distribution of noiseexposure. Thisview,developed underthe sponsorship of the U.S.Air ForceNoiseandSonicBoomImpactTechnologyprogram,canassignexplicitrolesto bothacousticandnonacoustic variablesin a probabilisticdecision-making process.The model,expressible in spreadsheet form,permitsidentification of twocomponents of annoyance judgments: directannoyance, basedonsensitivity totheimmediatecharacteristics ofacoustic signals, andresponse bias.Themodelalso providesinsightintotheextentto whichacousticvariablesmaybeconsidered to causeannoyance. UM AUDITORIUM, 9:00 A.M. TO 12:00 NOON SessionAAA. SpeechCommunicationVII: CrossLanguageStudiesof Production GeorgeD. Allen, Chairman Department of •4udiology andSpeech Sciences, PurdueUniversity, WestLafayette,Indiana47907 ContributedPapers AAA!. The coordinationof glottal with oral articulafiohs in Icelandic and Hindi.$ohn Kingston(DMLL, MorrillHall, CornellUniversity,Ithaca, NY 14853-4701) In mostconsonants producedwith glottalabduction,the openingis largelyor entirelycontainedwithintheaccompanying oral articulation. Furthermore, thepeakglottalopening occursat a constant intervalfrom the beginningof the oral articulationin unaspiratedstopsand voiceless fricativesand at a constantintervalfrom the releasein aspiratedstops, indicatingtightcoordination with theoral articulation[A. I_fifqvistand H. Yoshioka,"lnterarticulator programmingin obstruentproduction," Phonetica 3g,21-34 ( 1981) 1.IcelandicandHindipossess stopsin which the glottal openingtakesplacelargely outsidethe oral closure;in Icelan- dic presspiratedstops,abductionprecedesthe closureand in Hindi breathyvoicedstops,abductionfollowsthe closure.Both languages also have(post) aspiratedstopswith whichthe stopsof interestcanbe compared.Thisstudyexamined whetherexternalglottalarticulations arealso vary with the precedingstop closure.In Hindi, the duration of both breathyvoiceand (post) aspirationcoratied with the durationof the preceding stopclosurebutthedurationof neithercovariedwiththatof the followingvowel.BothkindsofHindistopsarelikethoseobserved in other languages, but the relativetimingof glottalandoral articulations in the Icelandicstopsis dearly different.In that language,the oral articulation with whichthe glottalarticulationiscoordinatedisthe vowelor the vowel and consonant combined. 9:15 AAA2. Temporal effectsof geminateconsonantsand consonantclusters. MargaretH. Dunn (Yale Universityand HaskinsLaboratories, 270 Crown Street, New Haven, •F 06511 ) Current phonologicaltheory analyzesgeminateconsonantsas quences ofadjacenttimingslotsthatcompletely sharefeatures whileclosingonesyllableandopeningthenext.Thisanalysis predictsthatthetem- tightlycoordinated with theoral articulation of the consonant. Onefe- poralorganization of utterances withgeminate consonants isparallelto malespeakerof eachlanguagehasbeenrecordedsofar, producingwords containingthe stopsof interestat differentratesand with stresseitheron thesyllablecontaining thestopor not.If theoralandglottalarticulations are coordinatedwith one another,then changesin their individualdurationsdueto changes in rateor stresslocationshouldcoysty.The duration of bothpre-andpost-aspiration in Icelandiccovariedwith thedurationof the adjacentvowel,and the durationof preaspiration alsocovariedwith the followingstopclosure,but the durationof postaspiration did not co- that of utterances includingheterosyllabic consonant clusters.This predictionwas testedusingFinnishand Italian, two languageswith very differenttemporalorganizations. Five Italian andfour Finnishsubjects S114 J. Acoust. Soc.Am.Suppl.1, VoL82, Fall1987 readlistsof nonsense wordsincludingbilabialgeminatcsandbothhomor- ganicandheterorganic clusters. The hypothesized equivalence of geminatesand clusterswastestedby examiningthe amountof closedsyllable vowelshortening exhibitedin thetwoenvironments. The resultsdid show that geminatespatternedwith dusters,with respectto precedingvowel 114thMeeting: Acoustical Societyof America S114 durationin bothlanguages, butclosed syllablevowelshortening wasexhibitedonlyin theItalian.This •sult suggests that the syllablemaybe usedasa unitof temporalorganization in Italian,butsomeotherunitsof temporalorganization mustbe proposed for Finnish.[Work supported by NIH Grant NS-13617.] casualspeaking styles.The paperis particularlyconcerned with output constraints that canbe understood in termsof systems of phonemiccontrast.Specifically, it ispredicted thatwhenthephoneroes ofa language are widelyspreadout in the acousticspace,then the acousticoutputconstraintson them are lessstringentthan if the phoneroes werecrowded together.Data supportingthis view will be presentedfor vowel-to-vowel coarticulation in threeBantulanguages.[Work supportedby NIH to 9:30 HaskinsLaboratoriesand NIH to M. I. T. ] AAA3. VOT variability in stop-consonantproductions by bilingual speakersof Englishand Mandarin Chinese,Judith L. Lauter and Fang Ling Lu (Departmentof SpeechandHearingSciences, Universityof 10:15 Arizona, Tucson,AZ 85721) It hasbeenobserved[e.g.,L. Williams,Percept.Psychophys. 21,289297 (1977) ] that somebilingualspeakersof Spanishand Englishwill use language-appropriate VOT valuesin elicitedsetsof Englishand Spanish stops.As partof an ongoingcross-language projectin repeated-measures acousticphonetics, six nativespeakers of MandarinChinese(three femalesandthreemales)wereaskedto producerepetitions of Englishand Chinesestopsin a nonsense/ha-'Ca/context. For theChinesestops,five separateserieswerecollected,with the stressed syllablepronouncedon eachof fiveChine• tones.Productions weredigitizedviapulsecodemoduiator,storedon videotape,and VOT segments identifiedand measured usingwaveformanalysissoftwaredeveloped in our laboratory.Measurementsof meanVOTs for eachtalkerindicatethat Englishand Mandarin Chineseare similarin rangeof absoluteVOT values,clearVOT distinctionsbetweenvoiced/voiceless stops,and inconsistent useof VOT to distinguishplaceof articulation.Further,VOT oariabilitypatternssuggest that for someof thesetalkers,controlof VOT timing may be sensitiveto tonalcontext,and thissensitivitymay be reflectedin productionsof Englishaswell asChinesestops.[Work supportedby AFOSR.] 9:4• AAA6. Supportfor the frame modelin Turkish and in English.Suzanne E. Boyee(HaskinsLaboratories, New Haven,CT 06510) The look-ahead and the framemodelof anticipatorycoarticulation makedifferentpredictions aboutlip movementin a wordsuchas/ustu/. The look-aheadmodelpredictsthat roundingfromthesecondvowelwill spreadinto the intervocalicconsonants, producinga singlemovement with a long sustainedphase;the frame modelpredictstwo distinctgesturesfor theroundedvowelswhosetailsmayoverlapsomewhat, producingtwopeakswitha "trough"between. In earlierwork,it wasshownthat for nonsense•vords, such as /kuktluk/, English speakersproduce troughs,while Turkish speakersshow plateaulikepatterns.Does this meanthatTurkishandEnglishspeakers employ,respectively, look-ahead and framestrategies for eoarticulalion? Alternatively,the framemodel holdsfor Turkish, but Turkish roundinggesturesare larger and longer, overlappingfor a greaterproportionof their tails.An additivemodelof overlapcouldthen producea plateaulikepattern.Speakers'movement tracesfrom/kiktlik/and/kuktluk/were addedandthe resultcompared with thosefor/kuktluk/; a goodfit for Turkish speakerswastakenas evidencefor theframemodel.Further,theproportionof roundingduring the consonantintervalwas differentfor the two languages.Both tests supporttheframemodal.[ Worksupported by NIH. ] AAA4. Stop consonant voicing in Malay-English bilinguals. Grace H. Yeni-Komshianand Inderjit K. Bhathal (Departmentof Hearing andSpeechSciences, Universityof Maryland,CollegePark, MD 20742} The productions of two groupsof Malay-Englishbilingualswereinvestigated. The subjects, nativespeakers of Malay, wereaskedto read Malay and Englishwordsthat differedin voicingof initialand finalstop consonants. Thefoursubjects in theMalaysia groupresided in Malaysh andwerenotexposed to a predominantly Englishspeaking environment, and the two subjectsin the U.S. groupresidedin the U.S. for 5 years. Resultson word initial stopsare reportedhere.In Malay, voicedinitial stopsarecharacterized by voicingleadandvoiceless stopsby shortvoicing lag.Comparison of the twobilingualgroupsrevealedthat, although therewerenosignificant differences in thevoicedproductions, therewasa shifttowardsa moreEnglishlikeproductionfor the voiceless stopsby the U.S. group.Analysisof the Englishwordsshowedno differences in the productionof the voicedstops;however,the U.S. groupproducedthe voiceless stopswith significantly longerrOT valuesthan the Malaysic group.Noneof the subjects showeda shiftin voicingcharacteristics asa functionoflanguage. Theseresultssuggest thatnativeMalayproductions in bilinguals,residingin theU.S., werealteredin a directionthat ischaracteristicofEnglish. 10:30 AAA7. Quantitativecharaeterizationof degreeof coarticulationin CV tokens.Abigail C. Cohn (Departmentof Linguistics,Universityof California,Los Angeles,CA 90024-1543) Languages areoftenassumed to showgreateror lesserdegrees ofcoar- ticulation[ P. Ladefoged, in Papers in Linguistics andPhonetics in MemoryofPierreDelattre,editedbyA. Waldman(Mouton,TheHague,1972), pp.273-286], yetnoquantitative measure existsforcharacterizing degree of coarticulation. In thisstudy,thedevelopment of a metricfor degreeof influenceof a vowelon a precedingstopconsonant hasbeenundertaken. For eachlanguage in thestudy,LPC spectraweremadeat theburstonset of initialstopsfollowedbya vowel.Templates weremadeof thesespectra whenthe vowelwas/a/. The stopspectrawith othervowelsweretested againstthesetemplatesto determinea percentfit. Languages differin theirpercentfit;a lowpercenttit indicates that vowelshavea greateffect onpreceding consonants anda highpercentfit indicates thattheydonot. Kana,a language of Nigeriawhichgivestheauditoryimpression of havinga highdegreeof coarticulation, is beingcompared to otherlanguages includingRussian,a language with secondary articulations. [Work supportedby NSF.] 10.'00 AAA•. Output enna•raints and cross-languagedifferences in coarticulation. Sharon Y. Manuel (Bldg. 36-529, Massachusetts Instituteof Technology, Cambridge, MA 02139) Coarticulatory patternshavebeenshownto varyfromlanguage to language, fromspeaker tospeaker, andfromonestyleor rateof speaking tothenext.Manyofthese differences mightbeexplained byassuming ( 1] thatphones areassociated withranges of acceptability, thatis,"acoustic outputconstraints," whichmust not be violatedby coarticulation-indueedvariabilityand (2) that theseoutputconstraints vary from languageto language, speakerto speaker,etc. For thisnotionto haveany predictioe value,it is necessary, in turn, to be ableto predict,to some extent, justwhattheacoustic outputconstraints are,andhowtheymight vary.It mightbeexpected, for example,thatoutputconstraints relaxin S115 ,l.Acoust. Soc.Am.Suppl. 1.VoL82.Fall1987 10:4• AAAg. Timing of contextualnasalizationin two languages.Marie K. Huffman (Departmentof Linguistics,Universityof California,Los Angeles,CA 90024-i 543) Coarticulatory patternsin nasalityin twoWestAfricanlanguages--AkanandEfik will bedescribed. Theselanguages makedifferentphoriologicaluseof the feature{nasal). Nasalityisdeterminedfrom nasaland oralairflowasrecorded witha Rothenburg splitmask.Thefacusisonthe timingof contextual nasalization; in particular,thecoarticulatory effects of nasalconsonants onadjacentvowelsthatarenotcontrastively ( q- nasal}.Theseeffects aresaidto varydepending on thephonologieal useof nasalityin a language.In Akan, thereisan oral/nasalcontrastonvowels, 114thMeeting: Acoustical Society ofAmerica $115 evenin theenvironment of nasalconsonants; thuscontextualnasalization ized like the lax vowelsin Germaniclanguages, but couldstill be de- maybelimited,soasto preserve thephonemic contrast. In Efik,thereis scribed with a feature lax. no suchcontraston vowels,socontextualnasalizationof vowelsmay be moreextensive, beginning earlier,andendinglater,ashasbeenreported elsewhere forEnglish. Dataforseveral speakers ofeachlanguage willbe 11:30 discussed. [Work supported by NSF.] AAAII. Cross language study of the effects of voiced/voiceless consonants on the vowelvoicesourcecharacteristics. AilbheNi Chasaideand Christer Gobl (Centrefor LanguageandCommunication Studies, TrinityCollege, Dublin2, •.ireandDepartment of Speech 11:oo AAA9. Cross-linguisticarticulatory models of vowels. Michel T. T. Jackson (Phonetics Laboratory, Department of Linguistics, Universityof California,LosAngeles,CA 90024-1543) Themathematical andphoneticfoundations of a familyof techniques for quantitatively analyzingarticulatorpositions duringspeechproduction will be discussed in termsof a smallnumberof modesthat spanthe observed spaceof articulatoryvariation.Particulararticulatorypositions arerepresented asweighted combinations of themodes; variousnaturalhessconditions that dependon havingmultispeaker datacanbeusedto resolve mathematical indeterminacies in the choice of modes. Results us- ingmeasurement of articulatorpositions fromsagittalx raysof English, Icelandic,Spanish, andAkanvowelssuggest thatthebasisneededfor any onelanguage isof relativelysmalldimensionality. Typically,onlytwo or threemodessuffice to generate theobserved rangeofarticulatorpositions. However,not all languages havethesamemodesof articulatorpositioning.Thusa cross-language modalanalysis requires higherdimensionality, i.e., moremodes.In conclusion,thereare systematicdifferences between languages in thefunctionalorganization of a setof articulatoryprimes. Thesedifferences in functionalorganization !endto the observed crosslanguage differences in modalorganization. [Worksupported by NSF.] Communication and Music Acoustics, KTH, Box 70014, S-100 44 Stockholm, Sweden) Sourcecharacteristics of a vowelmaydifferaccordingto the voiced/ voiceless natureofadjacentconsonants. Thepostvocalie consonant could beparticularly crucial,asvocalfoldabduction for a voiceless consonant maybeinitiatedconsiderably beforeoralocculsion. Here,CV (:) C utterances(whereC = voiced/voiceless labialstopor fricative)wereanalyzed forfemaleandmalespeakers ofSwedish, English,andFrench.Characteristicsof the voicesourceweremeasuredfrom inversefiltereddata using thefour-parameter LF modeldeveloped at theKTH. Airflowrecordings witha Rothenberg maskallowedinferences regarding incomplete vocal foldclosureduringthe vowelandat onsetandoffsets. Resultsindicate majoreffects ofa following consonant; thelaterpartof thevowelprecedinga voiceless consonant shows a markeddropinexcitation strength and a steeper spectral slope,asmightbeexpected whenthevocalfoldsare opening butvibrating. A spectral consequence ofthisabducting gesture is a widening of theF 1bandwidth andan upwardshiftin formantfrequencies. These effectswere much less in evidence in French than in Swedish andEnglish. Thepreceding consonant hadcomparatively littleeffect; full excitationstrengthis achievedalmostimmediately. 11:45 11:15 AAAI0. Acoustic correlates of TENSE-LAX vowels and ATR vowels. AAAI2. An investigationof tag intonation in English. Richard W. SproatandShirleyA. Steele(AT&T BellLaboratories, 2D-443,600 Mona Lindau-Webb (PhoneticsLaboratory, University of California, LosAngeles,CA 90024-1543) Mountain Avenue,Murray Hill, NJ 07974) Germanicvowelsystems typicallyincludetwo setsof vowels,a long, tensesetanda short,lax set.Formantchartsof English,German,Dutch, and Swedishshowthat lax high and mid vowelstend to have a more centralizedpositionthantheir tensecounterparts.Tenseandlax low vowels do not showany suchsystematicrelationships in the differentlanguages. In Africanlanguages with vowelharmony,theharmonizing sets aresometimes labeledtense-lax,althougha moreappropriatelabelis advanced-retractedtongueroot (ATR). Formant chartsof Akan, DhoLuo, and Agwagwune(a Nigerianlanguage)werecomparedto the Germanicones.The tense-laxand + / - ATR featuresbehaveverysimilarly "'Are youcoming?'hesaid,"where"he said"isan attributivetag,and"I Tagsfrequentlyoccurin Englishtext. Commonexamples include boughta newcar,Manny,"where"Manny"isa vocativc tag.Beenuse of thefrequency of tags,it is importantto havean adequate treatmentof theirintonationfor morenaturalsoundingtext-to-speech synthesis. This paperexamines theintonational properties of tagsin AmericanEnglish and discusses their phonological characterization. Investigatedinstrumentallyare the F0 patternsemployedby speakers producingtagsin question andstatement contexts("'Are youhungry?' hesaid"versus "'I amhungry,'hesaid").A phonological analysis isproposed forsuchtags buildingon workby Pierrehumbert ["Thephonology andphonetics of for front vowels.Back - ATR vowels,however,tend to havea lower F2 than their + ATR counterparts,makingthemmoreperipheral,not more Englishintonation,"Ph.D. dissertation, MIT (1980) ] andBeckmanand Pierrehumbert ["Intonationalstructurein Japanese andEnglish,"Pho- centre] like back lax vowels. Thus tense and ATR are not the same acous- nologyYearbook 3, 255-309(1986)]. Finallytheincorporation of these tically. Althoughthesetwo featuresdo not occurcontrastivelyin any language,theydo possibly bothoccurin ruleswithina language.Agwagwunehascentralizedallophones in closedsyllables. Thesearenotcentral- resultsintothetext-to-speech systemunderdevelopment at BellLaboratoriesisdiscussed, andtheresulting improvement in thequalityofsynthe- $116 J.Acoust. Soc.Am.Suppl. 1,Vol.82,Fall1987 sis is demonstrated. 114thMeeting: Acoustical Society ofAmerica $116 FRIDAY AFTERNOON,20 NOVEMBER1987 UNIVERSITY LECTUREHALL, 1:30TO 4:30P.M. Session BBB.Psychological andPhysiological Acoustics IX: Physiological Acoustics •zcan•zdamar,Chairman Department of Biomedical Engineering, Unioersity of Miami, CoralGables, Florida33124 ContributedPapers 1:30 BBB1. Phuse-lockof spontaneousoto-acousticemissionsto a cubic difference tone.Pim vanDijk andHeroP. Wit (InstituteofAudiology, P.O. Box30.001,9700RB Groningen,The Netherlands) at 0.5 kHz, 95 dBSPL,eachfor9 days.ThisSOAE,notpresentbeforethe secondexposure,wasa narrow-bandsignalat 2.2 kHz that variedfrom 10-26 dB SPL acrossrecordingsessions. The SOAE hasremainedstable overthe3-yearperiodsinceit wasdiscovered. Duringthepastyear,this signal. Therefore, twotoneswithfrequenciesf• andf2 andlevels L• and animalhasbeentrainedbehaviorallysothat detectionanddiscrimination thresholds canbe determinedin the regionnearthe SOAE. Measuresof auditorysensitivityat quarter-octave intervalsbetween0.125 and 16.0 kHz revealeda hearingloss(re: normalchinchilla)of 22 dB centeredat werepresented to earshavinga spontaneous oto-acoustic emission. Here, 2.37 kH2. However, thresholdsdetermined at 10-Hz intervals between f• andf2 produced a 2f• --f2 distortionproductcloseto ( < 50 Hz) the emission frequency fo. Frequency spectraandzero-crossing time trend recordings wereobtained forvariousvalues ofœ•,L2,fl, andf2. Whenthe L• andL• weresufficiently high,the emission wasphaselockedto the 2f• -f• tonegenerated in theear.WhenL, andL 2weresmall,theemissionwas:mstained atfo. At intermediate levels,the emission is not constantly1o:kedto 2f• -f•, butoccasionally jumpstofo.Thebehaviorde.- 1600Hz and2800 Hz showedthat the SOAE wascorrelatedgenerally with a regionof enhancedsensitivityin the microstructure of the audiogram. However,in the immediatevicinityof the SOAE thresholdswere elevatedby 5-10 dB;additionof a tonalmasker(2.2 kH2 at 5-20 dB SPL) If a singleexternaltoneisusedto studyphase-lock of a spontaneous oto-aeoustie emission, it isdifficulttoseparate thistonefromtheemission scribed above is identical to that of a self-sustainedoscillator in the, presence ofnoise, drivenbyanexternal tone[R. L. Stratonovich, Topics in the Theoryof RandomNoise(GordonandBreach,New York, 1963), Vol. II, Chap.9]. [Work supported by Z. W. O.] hadnoeffecton theshapeof theaudiogram. [Worksupported by Grant OH 02128 from NIOSH. ] 2:15 BBB4. Interaural phase and level effects on binaural interaction component (BIC) of the brain-stem auditory evoked potential. BallachandaB. Bopannaand GeorgeMoushegian(Callier Center UTD, 1966 Inwood Road, Dallas, TX 75235) 1:45 BBB2. Effectsof frequencyseparationof primary toneson the amplitude of acousticdistortion products.Frances P. Harris (Department of Speech andHearingSciences, UniversityofArizona,Tucson,AZ 85721), BardenB. Stagnet,Glen K. Martin, and Brenda L. Lonsbury-Martin (Department of Otorhinolaryngologyand CommunicativeSciences, BaylorCollegeof Medicine,Houston,TX 77030) The effectsof frequency separation of equilevelprimarytones, f• and f> on theamplitudeof distortion-product emissions (DPEs) at 2f• -f2 wasinvestigated in 10earsoffivehumanswithnormalhearing.TheDPEs at 1,2.5,and4 kHz in response to primariesat 65, 75, and85dB SPL were: generatedwithf•/ft ratiosvaryingin 0.02 stepsfrom 1.01to either 1.41 (4 kHz), 1.5'9(2.5 kHz), or 1.79( 1kHz). Resultsindicatedthat, in generat.. ingthemaximumDPE amplitude, fe -f• frequency separation wassys.. tcmatieallyrelatedto stimulusfrequencyandlevel.Thus,for all levelsof stimulation, f•/f• ratio wasinverselyrelatedto DPE frequencyso that higherratiosweremosteffectivein generatingDPEs at I kHz, while lower ratioselicitedlargeemissions at 2.5 and 4 kHz. In addition,for all three DPiEs,in obtainingmaximumresponses, primary-tonelevel was directlyrelatedtof•/ft ratio in that lowerratioswere mostefficacious in generatingDPEs to 65-dB stimuli,while higherratioselicitedlarger DPEs in response to 75- and 85-dBprimaries.On the average,however,. across all frequency-amplitude stimulus conditions, maximalDPE ampli-. tudewashestobtainedusinganf:/f• ratio of 1.23.Overall,DPE findings describing the outcomeof varyinglevel,frequencyseparation, and fre-. quencyregionof primarytoneswereconsistent with thosereportedin the psyehophysieal literaturefor normal-hearing humans.[Work supported by NINCDS. ] 2:00 Electrophysiological studieshaveelegantlydemonstrated that neuronsin lowerbrain-stem nucleiareexquisitely sensitive to theparameters of soundthat mediatelocalizationandlateralization phenomena. In the presentstudy,thebinauralinteractioncomponent(BIC), derivedfrom brain-stem auditoryevoked potentials, wasobtained usinglow-frequency stimuli with various combinations of time and level differencesdelivered throughearphones. Theaimhasbeentodetermine whetherthesevolumeconducted multineuronal responses revealanyofthe neuralmechanisms of lateralization. The findingsshowthat characteristics of the binaural interaction component, e.g.,latency,amplitude, andwaveform morphology,evokedby low-frequency tonebursts,aredifferentially alteredwith changes in interauralphase.The effectson thebinauralinteractioncomponentwith interauralleveldifferences werealsosubstantialbut not similar to thechanges producedwith phase.From thisresearch, it hasbecome apparentthat,givenappropriate recording andstimulating parameters, manynuances ofbinauralhearingmaybeneurophysiologically addressed in studiesof volume-conducted evokedandderivedpotentials. 2:30 BBBS. AP and ABR tuningcurves:A comparativestudy.Abdclhamid A. Elshintinawy and Paul J. Abbas (The University of Iowa, Departmentof SpeechPathologyand Audiology,Wendell Johnson Speechand HearingCenter,Iowa City, IA 52242) Whole-nerve actionpotential(AP) andauditorybrain-stem response (ABR) wererecordedfrom anesthetized eats.Response amplitudeand latencyto a probestimulusweremeasured andtuningcurveswereconstructedusingthreemaskingparadigms; forward,simultaneous (a tone burst maskeroverlappingthe probe), and continuous(a continuous maskertone). AP andABR tuningcurvesweregeneratedon thebasisof criterion changesin both amplitude and latency of the response.Within BBB3. Spontaneous oto-acousti½ emissionfroma chinchillaear following exposureto noise.William W. Clark and Martha Solomonson(Central Instituteibr theDeaf, 818 S. Euclid,SaintLouis,MO 63110) eachof themaskingtechniques, tuningcurvesdeveloped fromtheampli- A spontaneous oto-acousticemission(SOAE) was observedin one chinchillaaftertwoexposures to anoctavebandof noise(OBN) centered significantdifferences. The only significantdifferences were measured acrossmaskingconditions. Forwardmaskingtuningcurveswerefoundto havesteeperhigh-fre•quency slopes, higherQ 10values,andlowersensitiv- Sl 17 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 tude and latencycriterion changesof the AP and the ABR showedno 114thMeeting:Acoustical Societyof America $117 ity indicies(definedasthe difference betweenprobeleveland the levelat the tip of the tuningcurve). Low-frequency slopesandthe tip-to-tailratiosshowedno significantdifferences. [Work supportedpartiallyby the DeafnessResearchFoundation.] 3:30 BBI]9. Evalust'onof aneight-channelscalatympanielectrodewith single unit recordingin the auditory nerve. David. H. Liang (AEL 128. Departmentof ElectricalEngineering, StanfordUniversity,Stanford,CA 94305) 2:45 •zcan •zdamar, RafaelE. Delgado(Department of Biomedical In orderfor a multichannelcochlearprosthesis to be moreeffective thana single-channel prosthesis, thedifferentchannels of thestimulating electrode mustbeableto activateseparatepopulations of theneuronsin Engineering,Universityof Miami, Coral Gables, FL 33124), and Ferdinando Grandori (Polytechnicof Milan, Milan, Italy) the auditory nerve. Earlier animal studies[C. van den Honert and P. Stypulkowski, Hear. Res.14, 225-243 ( 1984);H. S. Lusted,Ph.D. thesis, BBB6. Fiber tract model of auditory brain-stem responsegeneration. A fiber tract model for the generationof auditory hrain-stemresponses (ABRs) is developedusingvolumeconductedpotentialsin a sphericalconductingmedium.This modelproposes that the fastcomponeatsof theABR aregenerated by the synchronized propagation of the actionpotentialstravelingalongthe auditorybrain-stempathways.The modelrepresents anactionpotentialwith upto 100currentdipolestravelingalonga pathwaycorresponding to a knownauditorybrain-stemtract. Coordinatesof the majorfibertractsare obtainedfrom hamanbrainatlasesand adaptedto a sphericalhead model.Computationsassuming classical interconnections between brain-stem nuclei show that such synchronously propagatingaction potentialscan generatewaveforms similarto recordedABRs. Accordingto thismodel,positivepeakscorrespondtothedepolarizing wavefrontof theactionpotentialwhilenegative valleysare generatedwith the repolarizingphase. StanfordUniversity,1980]haveresultedin conflicting conclusions about whetherthisoccursfor monopolar stimulation. To address thisquestion, an eight-channel electrode arraywasacutelyplacedin thescalatympani of sevencats.Recordings werethenmadein the VIIIth nerve,of single unit responses to electricalstimulationfrom eachof the channelsin the array.For eachunit isolated,thresholdcurrentfor hipbasicmonopolar stimulation was determined for each of the electrode channels. Some units werefoundtobehighlyselective intheirresponse tothedifferent channels of theelectrodearray,with themosteffectiveelectrodehavinga threshold lowerthan an adjacentelectrodeby up to a factorof 5 times.Adjacent electrodes wereeither 1.0or 1.5mm apart.Other unitswerefoundnot to beveryselective in theirresponse. Thissuggests thatsomedegreeof spatial selectivity isachievable withmonopolar stimulation.[ Worksupported by NIH.] 3:45 BBBI0. E!ectromagnet'cbasis of acoustic perception. I. Biological BBB7. Steady-stateevokedpotentialsin sleepinghumansto continuous modulated tones. Lawrence T. Cohen and Field W. Rickurals structures.Glenn M. Cohen (Department of BiologicalSciences, FloridaInstituteof Technology, Melbourne,FL 32901) (Departmentof Otolaryngology, Universityof Melbourne,Parkville, Australia 3052) At a previousmeetingof the society,Rickuralsand Clark [L Acoust. Soc.Am. Suppl.172, S54(1982)] showedthatsteady-state evokedpotentialscouldberecorded fromalerthumansin response toamplitude-modu- iatedtonesovera widerangeof modulation frequencies. A subsequent paperhasshownthat,whilethereisanamplituderesponse maximumfor alert subjectsat a modulationrate of approximately40 Hz and this frequencyis thus appropriatefor optimumthresholddetection,different circumstances applyfor sleepingsubjects. Sleepingsubjects differfrom alertsubjects in bothamplituderesponse andbackground EEG noise,as functionsof modulationfrequency.A detectionefficiencyfunctionre- vealsthatduringsleeptheoptimummodulation frequency, dependent on Althoughhearingaidscancompensate forsomehearingimpairments, theycannoteffectively compensate for impairedfunctions resulting from majorlosses of haircellsand/or nerves.Cochlearimplantsarelimitedin theireffectiveness bythenumberof functional cochlear neurons, surgical techniques, andcurrenttechnology. Electromagnetic waves,a newapproach,offerthepotentialofbypassing theseverely impairedcochleaand directlysimulating thehigherauditorycenters, suchastheauditorycortex [P. L. StocklinandB. F. Stoeklin,TIT J. Life Set.9, 29-51 (1979) ]. Thisapproach isbaseduponthefactthatexcitable tissues mayemitelectromagnetic wavesduringthedepolarization cycleduetoenergychanges in theintegralmembrane proteins. In thispaper,theauditoryapplication of electromagnetic wavesisintroduced byexaminingthebiological structuresimportantto acousticperception. stimuluscarrierfrequency,lieswell above40 Hz. For lowercarrierfrequencies a peakoccursat about90 Hz, whileat higherfrequencies the optimummodulation frequency isashighas 150Hz. Thesehighermodu- 4:00 lation rates allow more efficient detection than does 40 Hz. Stimuli with highmodulation rateshavebeenusedsuccessfully withsleeping subjects includingneonates. BBBI1. Electromagneticbasis of acoustic perception. II. Physical mechanisms.John G. Clark (Mentee, Inc., 8940 S.W. 129 Terrace, Miami, FL 33176) 3:15 BBBS. AGC behaviorof a recoverymodelfor auditory nearon firings. Timothy A. Wilson (Room 36-547, Department of Electrical Engineeringand ComputerScience,and ResearchLaboratoryof Electronics,Massachusetts Instituteof Technology,Cambridge,MA 02139} In a recoverymodel for auditory neuronfirings [ Westermanand Smith. J. Acoust. Sue. Am. gl, 680-691 (1987); Johnsonand Swami, J. Acoust.Soc.Am. 74, 493-501 ( 1983}], the firinghistorymanifestsitself as a reductionin the instantaneous probabilityof a neuralevent.In this paper,it is shownthat the ensemblebehaviorof sucha modelmay be interpretedas that of a feedbackautomaticgain control (AGC). The AGC input is the excitationfunctionand itsoutputis the ensemble-aver- agefiringrate.Theimpulseresponse ofthegain-control filterisa superpositionof time-invariantand time-varyingcomponents that are relatedto therecoveryfunction's absoluteandrelative-refractory intervals,respectively. In response to a steplikeinput, the modei'soutputresembles the firing-rateresponse of an auditoryneuronto a toneburst:Bothcontainan exponentiallydecayingrapid transientwhosetime constantdecreases as the input sizeincreases.[Work supportedby DARPA undercontract N00014-82-K-0727, monitoredthroughthe Officeof Naval Research.] Sl 18 J. Acoust. Soc. Am. $uppl. 1, VoL 82, Fall 1987 A studyof physicalparametersleadsto the conclusionthat the human skull,withbraintissueincluded,isa cavitythatwill supportelectromagneticstandingwavefieldsin the frequencyrangebetween200 MHz and 3 GHz. This frequencyrangecoincides with allowabletransitionsin the rotationalenergystatesof integralproteinsthat are embeddedin nerve membranes. Thisfrequency overlappermitstheintegralproteinsto couple energybetweenpropagatingactionpotentialsand standingelectromagneticfieldswithinthe skull.This isa crucialphysicalmechanism in a full theoryorbrain functioningwhich accordsstandingEM fieldsa central rolein acousticperception. All physicalmechanisms will bediscussed and relatedto biologicalstructuresin the brain. A key predictionof the theoryisthatthefrequenc•ofthelowestEM standingwavemodeforany individualcanbepredictedsimplywith knowledge of thecephaltoindex. Recentmeasurements made on sevenindividualsconfirmthe theory. [Work sponsored by Mentee,Inc. ] 4:15 BBBI2. Effect of acoustictone perceptionon several higher modesof brain-generatedelectromagneticwaves: Preliminary evidence.Pitilip 114th Meeting: AcousticalSociety of America Sl 18 L. Stocklin(Consulting physicist, 439BlueJayLane,SatelliteBeach,FL 32937) Brain-generated microwaves weredetectedandmeasured for eachof twosubjects for severalhighermodesin thefrequency rangeof the7th13thprimary modeof the adult humanbrain ( 1200-1900MHz). Differeneesin modeamplitudeof 5-8 dB weremeasuredwith a loud acoustic FRIDAY AFTERNOON, 20 NOVEMBER tonestimulusON andOFF. Highermodeamplitudechangeoccursimmediatelyupontonestimulusstartor stop.In orderto increase higher modepowerwellaboveambientnoise,a zeromode"pumping" procedure wasused.Intersubject andcontrolsubject comparisons shownoacoustic/ microwave crosstalk, eithermechanical or electrical. No modeactivity wasmeasureable byanysubject in themodefrequency regionofanyother subject, withor withoutzeromodepumping.[Worksponsored by Mentec, Inc. ] 1987 UM AUDITORIUM, 1:30 TO 4:05 P.M. Session CCC. SpeechCommunication VIII: CrossLanguageandOther Studiesof Perception D. Kimbrough Ollcr, Chairman Mailman Centerfor Child Development,Universityof Miami, P. 0. Box 016820,Miami, Florida 33101 Chairman's Introductian--l:30 ContributedPapers 1:35 CCCI. Franco-American differences in the discrimination of DA from GA, FrancoisSanton(CNRS, Marseille,France) and Bertram Scharf (CNRS andAuditoryPerception Laboratory, Northeastern University, ferredsomemurmurandacceptedbriefernasalizationin the vowel;and Portugueseresponses were intermediate.Preliminaryacousticdata from utterances in thethreelanguages arein accordwiththeseperceptual findings. Boston,MA 02115) 2:05 Wesynthesized DAs andGAs thatweresupposed to differonlyin the third formant, between2 and 3 kHz. Discriminationwas measuredin narrow-band noisemaskerscenteredon frequencies from 1 to 3.5 kHz. Thesign:d-to-noise ratiowassettoyieldnearly100%correct responses in theleasteffective masker(at 3.5kHz) andnearly50% in themosteffectivemasker(at 1 or 2.5 kHz). ThreeAmericanlisteners yieldedresults similarto tuningcurves.Discriminationwaspoorestin the2.5-kHz noise and improvedrapidly at lower and higher noisefrequencies. Three Frenchliisteners gavesimilarresults, butthreeothersgaveverydifferent results.The signal-to-noise ratios were.11 dB lower and discrimination waspoorestin the 1-kHz noise.(One Americanalsodiscriminatedat the lowersignal-to-noise ratiowhengivenfeedback. ) Analysisof the tapes revealeda low-frequency clickin theDAs butnotin theGAs. Americans did not usethis cue (without feedback).SomeFrenchalsodid not, but others did. Apparently, becausenatural American DAs differ from FrenchDAs, Frenchlistenersare morelikely to useartifactualcues. 1:50 CCC2.Perception ofvowelna•alization inVC contexts: A cross-language study. Kenneth N. Stevens(Room 36-517, Massachusetts Institute of Technology,Cambridge,MA 02139), Amfilia Andrede, and M. C•u Viana(Centrode Linguistica da Universidade de Lisboa,Av. 5 de Outubro,85-6,1000-Lisbon, Portugal) A seriesof stimulirangingbetween[tAt,x],[t7•t,], and [tAnt•] were synthesi:,.ed bysystematically manipulating ( 1) thedurationof nasalizationin thevowel[ n], (2) theamountofnasalization inthevowel,and( 3) thedurationof thenasalmurmurfollowingthevowel.The stimuliwere presented to nativespeakers of Portuguese, English,andFrench,which differwithrespect totheoccurrence ofnasalvowels in theirphonological systems. The listeners wereaskedto judge,for eachstimulus,( 1) the presence or absence of nasalization and (2) theadequacy of thestimulus as a natural utterancewith respectto its nasalization.The different lan- guageg•oupsgavesimilarresponses with regardto the presence or absence ofnasalization. However, judgments ofthenaturalness ofthestimtlli in the differentlanguages dependon the temporalcharacteristics. Frenchliisteners preferred a longerdurationof nasalization in thevowel andwereindifferent to thepresence of murmur;Englishlisteners pre- $119 J.Acoust. Soc.Am.Suppl.1,VoL82,Fall1987 CCC3.Acoustic-phonetic analysisof Spanish/I/. Miguelina Guiraoand Enriquc A. Rosso (Laboratorio dc InvestigacionesScnsoriales, CONICET, CC 53, 1453BuenosAires,Argentina) Acousticmeasurements to examinethe contributionof overlapping segments--anticipatory andcarryovercoarticulations--totheidentification of AmericanSpanish/1/wereperformed.The consonant wascombinedwith five vowelsin naturallyproducedCV, VC, VCV, and CCV syllables. Acousticprofilesthroughanalysis of formanttrajectories, fundamentalfrequency, intensityanddurationof vowel,consonant, andtheir overlapping segments wereobtained.In searching for thecorresponding criticalspectral features, digitalizedwaveforms weremodifiedbydeleting varioustemporalsegmentsin the syllable.The excisedsoundswere used as stimuli in identifications tests. Results indicate that the duration of the syllabicnucleivariesacross front,roedial,andbackvowelsandalsoin preand post-vocalic position.Segment/1/presentedin isolationis poorly identified.Bestidentifications areobtainedwhenthestimuliareintegratedbytheendoftheoverlapping portionandthebeginning ofthevowel.A tradingoffrelationship effectisobserved betweenthesetwosegments. 2:20 CCC4. On the identificationof Spanish voiced stops. Jorge A. Gurlekian,Miguelina Guirao,andHoracio Franco(Laboratoriode Investigaciones Sensoriales, CONICET, CC 53, 1453 BuenosAires, Argentina) Experimentswereconductedto examinethe interactionbetweenrelevantacousticfeaturesof Spanish/b,d,g/emittedwith/a/in CV syllables and VCV combinations with the first or second vowel with stress. Durationof steadyandtransitionalvowelsegments andtheclosuredurationweresystematically segmented at particulartimeintervals.The entire sequences andsplicedsegments werepresented to nativeSpanishspeakers for identification. Resultswerepresented asa ratiobetweendifferentclosuredurationsand the numberof roeslit periodsrequiredfor the identificationof thestop.Differentratioswereobtainedfor eachconsonantin the CV syllable.Whenthe closureportionwasremovedcompletely,the un- voicedcounterparts wereaccurately perceived. Thiswasnottruefor/g/, wheretheabsence of the burstavoidedthe shiftingto/k/. For the VCVs whentheclosureportionwasentirelyremovedandtheremaining parts 114thMeeting: Acoustical Society ofAmerica Sl 19 wererejoined, thestops--approximants in thiscontext--were stillidentified.Theseresultsshowedthat/b/and/d/identification wascuedby a voicedportionpreceding thetransitionof the followingvowel.Primary featuresfor thesesoundswerelocatedin the transitionedges•final and initial--whichoverlapped bothvowelandconsonant information. 2:35 CCCS. Training new fricative contrasts, Donald G. lamieson (Department of CommunicativeDisorders, University of Western Ontario, London, Ontario N6G IHI, Canada), David Morosan (Departmentof Psychology, Universityof BritishColumbia,Vancouver, Canada), and Susan Rvachew (Department of Psychology,University of Calgary,Calgary,AlbertaT2N IN4, Canada) A perceptualfadingtechnique[cf. Jamiesonand Morosan,Percept. Psychophys. 40, 205-215 (1986) ] wasusedto train two typesof listeners to perceivenewfricativecontrasts.One groupconsisted of youngadult francophones who had difficultyproducingand perceivingthe voiced/ voiceless "th" distinctionin English.With thisgroup,trainingusingthe fadingtechniqueimprovedidentificationaccuracyboth for the specific, synthetictargetsusedin training,andfor multiple,naturaltokensof these sounds,spokenby differentindividuals,bothmenandwomen.However, this learningwasquitespecific,as performance waspoorwhenlisteners weretestedwith the targetsoundsin new word positions,or when/d/ tokenswereusedin the distractorset.The secondgroupof listenerswere youngchildrenwho consistentlymispronounced at least one English fdcativesound,and had beendiagnosed ashavinga functionalarticulation disorder.A sizeableproportionof thesechildren,all of whomhad beenselectedon the basisof productiondifficulties,displayedatypical perceptualskills;for thesechildren,trainingproduceda measureable improvementin identificationaccuracy,and in someinstances, perceptual training (without explicitproductiontraining) alsoimprovedthe productionsof thetargetsounds.[ Work supported by NSERC AHFMR and HWC:NHRDP. ] theydosoby notingnative-language (LI) phones in theL2, or bynoting divergences from normsestablished for phonesin the L2. Englishsenteneesspokenby Chineseand nativeEnglishtalkerswerepresentedto threegroupsof listeners. The Chineselisteners, whospokeEnglishwith strongforeignaccents,were able to distinguishChinesefrom native talkers.Thoseexperienced in EnglishresemblednativeEnglishlisteners significantly morethanthosewhowererelativelyinexperienced, suggesting the experienced Chineselistenershad developedtacit knowledgeof howEnglishsentences "ought"to sound.Removingpauses fromthesentencesdid not affectforeignaccentscores,suggesting that "fluency"and "foreignaccent"representdifferentdimensionspertainingto L2 proficiency.Chinesetalkerswho learnedEnglishbeforethe ageof 12 years receivedsignificantly lowerscoresthan nativeEnglishtalkers.Learning an L2 before12 years•the ageoftenassociated with the endof a "critical period"•does not guaranteeaccent-freepronunciation.There was no differencein foreignaccentscoresobtainedfor adult Chineselearnersof Englishwhohadlivedin theUnitedStatesfor 1and5 years.This suggests that amountof (unaided) L2 experiencemay not improveL2 pronunciation, at leastbeyonda certainpoint. 3:20 CCC8. Perceptual segregationof concurrent vowels. Peter F. Assmann and QuentinSummerfield(MRC Institute of Hearing Research, UniversityPark,NottinghamNG7 2RD, England) The abilityto grouptogethertheacousticcomponents of a voiceand segregate them from competingvoicesis an importantskill in speech communication. To studytheauditoryprocesses involved,pairsof vowels werepresented simultaneously to listeners withnormalhearing.Identificationperformance improvedas the fundamental-frequency difference betweenthe membersof the pair increased.Two auditorymodelswere comparedto predictthis result.The compositewaveformwas passed throughanarrayofbandpass filterssimulating thecat'slimitedfrequency resolution.The "place" model estimatedthe two fundamentalsfrom peaksin the profileof the rmsoutputsof the filters.It thensampledthe 2:50 CCC6. The role of duration in the percel•tion of consonantswithin reducedsyllables. Dani•le Archambault (Universit6 de Montr6al and INRS-T616communications, 3 Place du Commerce, Nuns' Island, Montreal, QuebecH3E I H6, Canada) The roleof durationasa perceptualcueto the presence of two underlying consonants versusone is examinedin Montreal French. Due to deletionof the interveningphoneticsegments,two identicalsyllable-initial consonants may bejoined in onearticulationsuchthat, for example, two plosivesshowa singleburst.A seriesof utterancesin which these doubleconsonants and the corresponding singleonesappearin identical segmental contextswererecordedandthenpresented to listeners in perceptualtests(e.g., la plupart (mostof it)•/laplypar/• [lap:at] as opposedto la part (the part) •/lapar/• [lapar] ). To insurethat listeners basetheir judgmenton durationto distinguishbetweenthe utterances, theywerealsopresented with thesamesentences in which,byuseof signal processing, thedurations of theconsonants wereinverted,i.e.,thedouble consonant wasgiventherelativedurationof thesingleconsonant andvice versa.Resultsshowthat while a longdurationconsistentlyleadslisteners to recognizea doubleconsonant,a shortdurationdoesnot automatically promptthem to identifya singleconsonant.Resultsalsoindicatethat consonantduration mustbeconsideredin a contextlargerthan that of the syllable.[Researchdoneat the Universit6de Montr6al and supportedby National ResearchCouncilof Canadaand Qu6becGovernment.] 3:05 CCC7. Non-native speakers can gauge degree of foreign accent in English.JamesEmil Flege (Departmentof Bioeommunieation, UAB, UniversityStation,Birmingham,AL 35294) Previousresearchindicatedthat non-nativespeakers candetectforeignaccentin theirsecond language(L2), evenif theythemselves speakit with an accent.This studyusedintervalscalingto helpdeterminewhether S120 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 profileat integermultiplesofeachfundamental to reconstruct thespectra of the constituentvowels.The "place-time"modelestimatedthe fundamentalsfromdominantperiodicities in thewaveformsemergingfromthe filters.The amountof synchronization to eachfundamental wasplotted as a functionof filter centerfrequencyto reconstruct the spectra.The place-timemodelestimatedthe two fundamentalsmore accuratelythan the placemodel,and predictedidentification responses moreprecisely. This outcomeis compatiblewith the ideathat a temporalanalysisis involvedin the segregation of concurrentsounds. 3:35 CCC9. Dichotic du!fiex pertelation and modularity. Ignatius G. Mattingly (HaskinsLaboratories,270 Crown St., New Haven, CT 06511) Recentwork suggests that dichoticduplexperception(in whichthe sameacousticmaterialsimultaneously yieldsbotha phoneticand a nonphoneticpercept)occurswhenthe source-defining or source-localizing processes are givenecologicallyiraplausibleinformation,and provides furtherevidence for themodularityofspeech perception. IfF 3 andF I-F 2 of a synthetic syllable are presenteddichotically, with identical F0 con- tours,onehearsjust the expectedsyllableat somefixedpositionwithin the head.With a somewhat higherF0 contouronF 3 thanonF 1-F 2, onehears two voices,oneat eachear, both sayingthe syllable.But with a considerablyhigherF 0 contouronF 3, perception isduplex:onehearsa buzzat the F 3earandthesyllableat theF 1-F2 ear.And if, instead,thebasicpatternis modifiedbyshiftingF 3 abruptlyto the F 1-F 2 earafter80 msor more,one hearsjust the syllable,now moving acrossthe head. But if F 3 is shifted sooner,perceptionis againduplex:onehearsa chirp at the ear initially receivingF3 and the syllableat the otherear. Evidently,sufficientlyextremeconditionsdisruptsourcedefinitionor sourcelocalizationwithout disruptingdichoticspeechperception.Speechperceptionappearsto be independentof theseother modularprocesses, and thereforemodular itself. [Work supportedby NIH Grant HD-1994.] 114thMeeting:AcousticalSocietyof America S120 3:50 CCC10.Synthesis andperception of paragraph-level prosody.Kim E. A. SilveJ•an (Room 2C-440,AT&T Bell Laboratories, 600 Mountain Avenue,Murray Hill, NJ 07974) graph textswere constructed:In each text the overall meaning(e.g., wherea speaker's turn finished)depended on thelocationof oneparticular paragraphboundary.In a seriesof perception experiments, listeners heardoneor anotherversionofeachtextspoken bya formantsynthesizer, and answeredforced-choice comprehension questions. Overall results showed that(i) listeners doexploitsuchwide-domain prosod icorganizaDiscourse structuresystematically influences the prosodyof utter- ances. in thisstudy, units ofprosodic structure thatspan multiple sentences(prosodicparagraphs)weresynthesized by (i) temporarilyincreasing •lheoverallpitchrangeat thestartof a (textual)paragraph; (ii) graduallydecreasing the rangeovera fixedtime windowat the end;and (iii) lengthening the pauseat the boundary.Six ambiguous multipara- FRIDAY AFTERNOON, 20 NOVEMBER tion whencomprehending extendedpassages of connectedspeech,(ii) the structurecanbesignaledby FO alone,although(iii) the moreredundantlyit isencoded, thebetterit will beperceived. Suchprosodic structure alsoimprovesthenaturalness of synthetic speech. The resultsarguethat prosodyis hierarchicallyorganized,rather than being a strictly linear sequence of intonationalphrases. 1987 UMS 4 AND 5, 1:30 TO 5:00 P.M. Sessiion DDD. UnderwaterAcousticsX: BottomInteractingOceanAcousticsII (Pr6cis-PosterSession) GeorgeV. Frisk, Chairman Bigelow208, WoodsHole Oceanographh: Institution,WoodsHole, Massachusetts 02543 Chairman's Introduction--l:30 ContributedPapers Followingpresentationof the pr6cis,posterswill be on displayuntil 5:00 p.m. 1:35 1:43 DDD1. Computationof the Blot coefficients.B. Yavari and A. Bedford (Departmentof AerospaceEngineeringandEngineeringMechanics,The University of Texas, Austin, TX 78712) DDD3. Seabed characterization from normal incidence reflectivity. To apply the Biot theory to the studyof the interactionof acoustic waveswith oceansediments,the coefficients in the equationsmust be evaluated.The coefficients dependon the frequencyand the fabric,or microstn•cture, of the sediments. A methodfor evaluatingthe dragand virtualmasscoefficients hasrecentlybeendeveloped[A. Bedfordetal., J. Acoust.Soc.Am. 76, 1804-1809(1984) ]. The methodrequiressolving for themotionof thefluidin theporeswhenthesolidconstituent is subjectedto a spatiallyuniform,oscillatorymotion.The finiteelementmethod hasbeenusedto determinethefluid motion.The dragandvirtual mass coefficients have beendeterminedfor a variety of two-dimensionalpore geometries. It wasfoundthat thedragcoefficient is insensitive to thepore geometry,while the virtual masscoefficientis very sensitiveto the pore geometry.It wasdeterminedthat the coefficients are independent of the frequencyup to a characteristicvalueof dimensionless frequencyand exhibita bilineardependence on frequency.[ Work supportedby ONR. I The effects of seabed interaction on air-to-water sound transmission. David M. F. Chapman(Institute of Soundand Vibration Research, The University,Southampton SO95NH, UnitedKingdom) The presenceof a reflectingseabedaltersthe transmission of sound fromanair-bornesourcetoa water-borne receiver. Resonances in a planewave transmissioncoefficientcorrespondto excitationof the acoustic normalmodesin the waterlayer.Althoughthe air-bornesourceexcites thesamenormalmodesasa water-borne source,theamplitudeandphase of the excitation coefficients are different. There are some notable effects: the weakdependence of transmission lossuponsourceheight,the preferred excitationof high-ordermodes,and the 25 log R dependenceof transmission lossuponrangein the mode-stripping regime.The predictionsof a simplenormal-modetransmission lossmodelare presented, includingrealisticabsorption in theair andat theseabed.[ Work supported by the:U.K. Ministry of Defence,throughthe Royal Aircraft Establishmentat Farnborough. $121 Surficialvaluesof sedimentcompressional velocity,density,and rms interfaceroughness are estimated from thecoherentcomponent of normal incidencereflectivityin the frequencyrange 10-1000 kHz. The Kirchhoffapproximation is employedfor the scattering functionin the inversionalgorithm such that the coherentreflectioncoefficientis as- sumedto be dependenton the d•stributionof the bottomheightsand independent of theshapeof thebottom.Thefunctionalformof thereflectiondatasuggests thatthedistribution of bottomheightsfollowsa Laplaclanrather thanGaussianform for a varietyof sedimenttypes.Laboratory analysisof bottom core samplesprovidedthe meansfor quantitative evaluationof velocityand densitypredictions. 1:47 DDD4. A numerical treatment of the fluid/elastic interface under range- 1:39 DDD2. Charles W. Holland and Burlie A. Brunson (Planning Systems Incorporated,7925 WestparkDrive, McLean, VA 22102) J. Acoust. Sec. Am. Suppl. 1, Vol. 82, Fall 1987 dependent environment. Er-changShang(Yale University,Department of ComputerScience,New Haven,CT 06520) and Ding Lee (Naval UnderwaterSystemCenter,New London,CT 06320) The actualphysicalproblembecomesconsiderably difficultand complicatedwhena fluid mediumis interfacingwith an elasticmediumunder range-dependent environment. In thiscase,a setof three"continuityconditions"musthold: ( 1) the continuityof verticalcomponents of displacement, (2) the continuityof verticalcomponents of stress,and (3) the horizontalcomponents of stressmustvanishontheinterface.The implicit finite difference(IFD) scheme,usedto treat the fluid/fluid interface IS. T. McDaniel and D. Lee, J. Acoust. Soc. Am. 74, 855 (1982) ] can be extended to treat the fluid/elastic interface with some modifications. In elasticinterfaceundera range-dependent environment.An attractivefeatureof thisnumericalmodelis that it is readilyadaptableinto the existing IFD code without excessiveeffort. Somecomputationalproblemswill this present anumer proce is prop to tre th flu also be discussed. 114th Meeting: AcousticalSociety of America S121 1:$1 DDD5. Transmissionof sound and beams through a water-sediment interface. T. G. Muir and N. L. Lunde (SACLANT ASW Research Centre,I- 19026La Spezia,Italy) Thecomputations of SchmidtandJensen[J. Acoust.Soc.Am. Suppl. 1 76, SI0 (1984) ] are extendedto examinethe penetrationof sound beamsintoa sandanda mudsediment asa functionof frequency, grazing angie,beamsize,andpulseshape.The computations are madewith the SAFARI model,an exactnumericalsolutionto the waveequationin horizontallystratifiedenvironments.The effectsof sedimentattenuation andrigidityareaccounted for. Beamshapecontoursandlosscoefficients in transmission throughtheinterfacearepresented. The effectof interface transmission ontheintegrityofpropagating pulsewaveforms isdiscussed. It isshownthatdecreasing thefrequency to thelowkHz rangeenhances the penetrationinto rigid sediments at low grazingangles(belowthe criticalgrazingangle).However,thisresultsin a geometric dispersion effecton the transmittedwaveforms,which destroysthe coherenceof pulsedsignals. 1:55 DDD6. Bottom interactingpulsepropagationin the ocean.Michael D. Collins (Department of EngineeringSciencesand Applied Mathematics,NorthwesternUniversity,Evanston,IL 60201) The parabolicequationmethodwasfirstappliedto underwater sound propagation by F. D. Tappert.It hassincebeenrefinedandbenchmarked intoa valuable toolforfrequency domaincalculations. Development ofan equivalentnonlinearmethodin the time domainwasinitiatedby B. E. McDonald and W. A. Kuperman.A linear versionof this methodhas beendevelopedto handlebottominteraction.The newmodelallowsden- sityandlosstobefunctions ofdepth.Lossisaccounted forin theproper mannerfor typicaloceansediments. Lossper meterincreases linearly withfrequency. A newinitialfieldhasbeenderivedusingthemethodof suredabsorption values,that theyexhibita maximumat 1750-mdepth, andthatthebehaviorof thismaximumisconsistent withotherpublished measurements andtheories.Third, Biotestimates of shearwaveabsorption are presented.Finally, a physicalmodel for the contact factor is presented. 2:07 DDD9. Determinationof oceanbottom sound-speed profilesusing a modal inverse technique in a range-dependentshallow water environment.George V. Frisk, JamesF. Lynch, and $ubramaniam D. Rajah (Department of Ocean Engineering, Woods Hole OceanographicInstitution,WoodsHole, MA 02543) A modalinverse technique fordetermining ocean bottom sound-speed profilesis appliedto experimental data obtainedin a range-dependent shallowwaterenvironment. The methodconsists of towinga cw source awayfrommooredreceivers to obtaina well-sampled measurement of the complexpressureversusrange.The Hankel transformof the measured fieldthenyieldsthe depth-dependent Green'sfunctionversushorizontal wavenumber, whichcontains distinctpeaksat themodaleigenvalues. The eigenvalues are usedasinputdatato a perturbative inverseschemethat generatesthe sound-speed profileof the bottom.Data at 140and 220 Hz wereobtainedin May 1984in NantucketSound,MA in an areawherethe depthchanged abruptlyat aboutthemidpointof theexperimental track. An associated change in bottomproperties wasassumed, andtheconcept of localmodaleigenvalues yieldinglocalgeoacoustic parameters wasthen usedto determine separate modelsfor thetworegions.Excellentagreementbetween thepressure fieldscomputed forthisrange-dependent model usingadiabaticmodetheoryand the experimentally measuredfieldsis demonstrated. 2:11 impulse response obtained by it intoa cwresponse andcomparing with DDDI0. Modelanddatacomparison for bottombackscattering. Darrell R. Jackson (AppliedPhysics Laboratory, Collegeof OceanandFishery Sciences, Universityof Washington,Seattle,WA 98195) and Kevin B. Briggs(Naval OceanResearchand DevelopmentActivity, NSTL, normalmoderesults.[Work supported by ONR.] MS 39529) matchedasymptotics. Numericalmethodssuitedto linearacoustics have beenimplemented. Themethodhasbeenbenchmarked byconvolving an Acousticandgeoacoustic datafrom two siteshavebeencompared 1:59 DDD7. Acoustic transmission across the water/sediment interface measuredduring in situ experiments.Lawrence J. Satkowiak (Naval CoastalSystemCenter,Code4120, PanamaCity, FL 32407-5000) A series of acoustic measurements were made to measure the trans- missionof acousticenergyacrossthe water/sedimentinterface.The in situmeasurements utilizeda three-dimensional arrayof hydrophones as acoustic receivers buried in the sediment. The acoustic sources were mounted on a movable tower located on the seafloor. The transmission wasmeasured asa functionof grazingangle,frequency, andpulseshape for two distinctlydifferentsedimenttypes,namely,fine sandand soft mud.The experimental resultspresented will becomparedto predictions madeusinga simplemultilayeredmodelthat treatsthe bottomasa series of contiguous layerswith knownacoustical properties. with a model [D. R. Jacksonet al., J. Acoust.Soc.Am. 79, 1410-1422 (1986)] for bottombackscattering strengthat highfrequencies. At one site,thesediment wasa finesandwithprominent, directional ripples.At the other site,the sedimentwasa mixtureof clay, sand,and shell.The geoacoustic datawereusedtoobtainvaluesforthefollowingmodelinput parameters: sediment soundspeed,sediment massdensity,andthepower spectrum for bottommicrorelief. The modelalsorequiresa sediment volumescattering parameter, whichwasobtainedby fittingbackscattering data.Themodelgenerally gavegooddescriptions ofthelevelandgrazing angledependence of thebackscattering strength,andtheresultsindicate thatroughness scattering wasdominantat thesandysite,whilesediment volumescattering wasdominantat theclayeysite.At thesandysite,the microreliefspectrumshowedappreciableanisotropyand an unusually rapid falloff with increasingspatialfrequency.For this site, the back- scattering modeloverestimated boththedegree of anisotroPy (nonewas observed)and the levelof scatteringat the highestfrequencies. [Work supportedby ONT and NORDA. ] 2:03 2:15 DDD8. Modelingdepth-dependent absorptionin marinesediments.Paul R. Ogushwitz (76 Cedar Lake West, Denville, NJ 07834) Blotmodelestimates of absorption areparticularlysensitive to porosity and to the frame moduli. In previouswork [P. R. Ogushwitz,J. Acoust. Soc. Am. 77, 429-464 (1985)], the frame moduli from a self- consistent theoryof composites (SCT) wasestimated,and an empirical factor (the "contact factor") that scalesthe SCT estimatesin order to DDDI1. High temporal resolution observations of hi•h-frequcncy acoustic bottom echoes. Thomas L. Clarke, John R. Proni (Ocean AcousticsDivision, NOAA/AOML, 4301 RickenbackerCauseway, Miami, FL 33149), and Lloyd Huff (NOAA/NOS/C_&GS, 6001 ExecutiveBoulevard,Rockville, MD 20850) matchmeasuredwavespeedswasdetermined.It isshownthat the contact factoris relatedto grainpackingandcementation. Here, it is firstshown that Blot modelestimatesof soundspeedversusdepth do agreewith measurementsin Gulf Coast brine sands[ G. H. F. Gardner et al., Geophysics39, 770--780(1974) ], wherethe SCT estimateshavebeenscaled bytheappropriatecontactfactor.Second,it isshownthatthecorrespond- Acousticbottomechoesat frequencies from 10 to 380 kHz wererecordedfor a varietyof bottomtypesin southernChesapeake Bay.Short pulselengthsanddirectdigitalrecording allowedtemporalresolution of 100psto beachieved.Supportingmeasurements of sedimentcharacteristicsandbottomroughness allowedcomparison of themeasurements with theory.The fieldobservations werein goodagreement with the theory after corrections for the interactionof transducerbeampatternswith ingBlotestimates ofabsorption versus deptharewithintherangeofmea- bottomroughnessweremade. S122 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 114thMeeting:Acoustical Societyof America $122 2:19 DDDI2. High-resolution acoustic bottom roughnessmeasurementin supportof bottomechointeractionmodeling.W. P. Dammannßand C. A. Lauter (Ocean Acoustics Division, NOAA/AOML, 4301 RickcnbackcrCauseway,Miami, FL 33149) A high-resolution acousticbottomprofilerusingan extremelynarrow-beam,3-MHz echosounderwasdevelopedat the OceanAcoustic Division of NOAA/AOML. The device was used to measure bottom roughnv;sover a rangeof scalesfrom lessthan I cm to severalmeters. Roughness measurements weremadein the lowerChesapeake Bay area overmud, fine to mediumgrain sand,and coursegrain sand.The data produced wereusedto appraise theperformance of an acoustic echoformationmodelthatpredictstheeffectsof marinebottomcharacteristics on a reflectedacousticpulseenvelope.Major aspects of thedesignanduseof the system,procedures for processing generated data,and examplesof processed outputare presented. isa featureless, gentlyslopingplatformboundedby the8- and 16-mdepth contours.Sedimentson this platformconsistalmostentirelyof mudsaccumulatedduringtheHolocenerisein sealevel.This areahasbeenexten- sivelysurveyedusing5- and 27-kHz seismicreflectionprofilingequipment.While generallyrevealingonly minor reflectorsin the mudswhich overliethepre-Holocene erosionsurface,certainareasconsisted of acousticallyimpenetrablesediments.X raysof corescollectedfrom theseareas showednumerousgas-filledvoidsin the sediment.Streamchannelmarginsare evidenton the pre-Holoceneerosionsurface,and the gaseous zonesare confinedbetweenthesemargins,althoughthe bottom of the channels cannotbeobserved dueto theacoustic impenetrability. Thegas zones are traceable and form an interconnectednetwork, one of which connectsto a presentday uplandstream. 2:35 DDD16. A comparisonof measuredbackscatterstrengthswith the firstand second-order terms of a Neumann series solution to the Helmholtz scattering integral. Richard Keiffer, M. F. Werby (NORDA Numerical Modeling Division, NSTL, MS 39529), and Steve Stanic 2:23 DDD13. Characterization of seafloortype and roughnessfrom 12-kHz acoustic backscattering measurements.C. de Moustier (Scripps Institution of Oceanography,A-005, La Jolla, CA 92093) and D. Alexandrou (Department of Electrical Engineering, Duke University,Durham, NC 27706) Seafloor acoustic backscatter measurements were carried out with 12- kHz seabeammultibeamecho-sounders overa varietyof terrainsincluding pelagicsediments on the top of two seamounts in the tropicalNorth Pacific;bemipelagic sediments, lava flows,and debrisflowson the East rift zoneof the Kilauea volcano;and a manganese nodulefield in the tropicalNortheasternPacific.Thesedata are subjectedto statisticaltests for homogcncity andstationarity,andsidelobeinterferencecancellationis performcdthroughadaptivefiltering.Thesemeasurements arcthcnuscd tocomparethe 12-kHzacoustic signature of theseterrainsbasedonstatis- (NORDA Ocean AcousticsDivision, NSTL, MS 39529) In previousASA meetings,a roughinterfacescatteringmodelunder developmentat NORDA hasbeenreported.This modelmakesuseof Brekhovskikh'stangentplaneapproximationfor the total field near the scatteringsurfacein the Helmholtzintegralequation.Direct numerical integrationis thenperformedyieldingthe first-orderterm of a Neumann series solution for the scattered field. Furthermore, as has been shown, this model also calculates a second-order term via Neumann's successive approximationtechnique.Physically,this second-order approximation beginsto accountfor secondaryscatterings as well as amelioratingthe initial tangentplaneapproximation.In thismeeting,the first-andsecondordermodeledresultswith backscatterstrengthsrecentlymeasuredoff of PanamaCity, FL, by NORDA's OceanAcousticsDivisionwill be compared. ticsoftheamplitude andthephase oftheechoes received atdiscrete angles of incidence in theinterval + 20'.Implicationsfor bottomtypeidentification aswell assurfaceroughness determinationarediscussed. [Research fundedby ONR. ] 2:39 DDD17. Correctingthe Kirchhoffroughboundary.Interactionmodelfor scattering.JohnJ. McCoy (The CatholicUniversityof America, Civil Engineering, Washington,DC 20064) 2:27 DDD14. Geoacousticpropagationthroughrandomseafloormodels.M. E. Dougberty and R. A. Stephen (Woods Hole Oceanographic Institution, Woods Hole, MA 02543 ) Finite differenceforwardmodelsof elasticwavepropagationthrough laterallyheterogeneous upperoceaniccrustare presented. The finitedifference tbrmulation is a two-dimcnsional solution to the elastic wave equationfor heterogeneous mediaand implicitlycalculatesP and SV propagation, compressional to shearconversion, interference effects, and interfacephenomena. Randomvelocityperturbations with Gaussianand self-similarautocorrelationfunctionsanddifferentcorrelationlengthsare presented thatshowdiffcrent characteristics ofsecondary scattering. The presence of a water-solidinterfacein themodelsallowsfor theexistence of secondaryStoneIcywaves,which accountfor much of the seafloor noiseseenin the syntheticseismograms for the laterally heterogeneous models. "Random" incoherentsecondaryscatteringincreasesas the productof wavenumber andcorrelationdistance(ka) approaches 1. Deterministicsecondary scatteringfrom largerheterogcneities is the dominant effectin the modelsas ka increasesabove 1. Secondaryscattering alsoshowsup asincoherence in the primarytracesof the seismograms when comparedto the laterallyhomogeneous case.Cross-correlation analysisof thc initial P-divingwavearrivalshowsthat, in general,the correlationbetweentracesdecreases aska approaches 1. 2:31 DDD15. Acoustic investigation of Palco-drainage networks in central ChesapeakeBay. Jeffrey Halka (Department of Natural Resources, Maryland Geological Survey, 2300 St. Paul Street, Baltimore, MD 21218) •tnd D. L. Gardner (NOAA, N/CGx3, Rockville, MD 20852) Locatedwestof the deepaxial channelin the centralChesapeake Bay S123 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall 1987 The Kirchhoffroughboundaryinteractionmodelis an incomplete largek approximation, whichdoesnot incorporate eithershadowing or multipleinteraction effects. Numericalimplementation suggests that,of thesetwo shortcomings, the neglectof shadowing is the moreimportant. Accordingly,attemptshavebeenmadeat correctingthe Kirchhoffmodel by introducing a shadowfunction.For statistical formulations, thisintroductionisfrequently accomplished aftertheincorporation of a statistical average. Thisis inconsistent withtheunderlying physics. Here,shadow functions areshownto havea rigorousbasisasstatistical measures of the roughboundarygeometry,independent of any scatteringexperiment. That is,onecandefinea suiteof statisticalmeasures of theroughboundary geometry,whichcansubsequently beinterpretedasshadowfunctions in the contextof a scatteringexperiment.The particularstatisticalmeasurethat is appropriatefor a gpecificexperimentis determinedby the statisticof the acousticsignalto be estimated. 2:43 DDD18. Soundscatteringfrom a randomlyroughfluid-solid interface. Dalcio K. Dacol and David H. Bennan (Code 5160, U.S. Naval ResearchLaboratory,Washington,DC 20375-5000) A generalformalismfor acousticscatteringfrom a fluid-elasticsolid interfaceisdeveloped. The theoryparallelsWaterman'swork [ P. C. Waterman, J. Acoust. Soc. Am. 45, 1417 (1969)], based on the extinction theorem, for acousticscatteringfrom inpenetrablesurfaces.This formal- roughfluid-solidinterfaceby usingperturbationtheoryto solvetheequationsfor the scatteringamplitudesto secondorder. The Watson-Keller renormalizationansatz [J. G. Watson and J. B. Keller, J. Acoust. Soc. Am. 74, 1887(1983) ] isthenusedto accountfor somemultiplescattering ism is then used to study the scatte of soun wav fro aral 114th Meeting:AcousticalSocietyof America $123 2:55 effects beyond double scattering. A numerical study ispresented comparingacoustic scattering in thefluidfroma fluid-solidinterfacewithacous- ticscattering in thesame fluidfroma hard,impenetrable surface (Neumannboundary condition).Substantial differences between thetwocases arefound. Besides theexpected acoustic energy lossbytransmission into thesolid, it wasalsofound thattheangular distribution oftheremaining DDD21. A $-MHz acoustic concentrationmeter that measures suspended sand. DanielM. Hanes(Division ofApplied MarinePhysics, Rosenstiel School, University ofMiami,4600Rickenbacker Causeway, Miami, FL 33149) acoustic energy thatisreflected andscattered bythefluid-solid interface A 3-MHz acoustic concentration meter(ACM), whichmeasures the differs greatly fromthehardimpenetrable surface case, especially near intensity ofbackscattered acoustic energy, hasbeen deployed inthelaboratory andinthefieldinorder toestimate theconcentration ofsuspended grazing incidence andscattering angles. sand.Techniques for measuring the profileof theconcentration of suspendedsandcloseto the seabedwith a verticalresolutionof about I cm 2:47 DDD19. Resonance interaction of sound waves with a stratified seabed. Greta Conde, P. D. Jackins, and G. Gaunaurd (Naval Surface WeaponsCenter,White Oak, SilverSpring,MD 20903-5000) A novelapproachto analyzethereflectionandtransmission of sound wavesfrom a penetrableand layeredfleetinterfacebetweentwo mediais described. Thisapproachgeneralizes thefirst-orderpredictionof theresonancescattering theory(RST), whichwasdeveloped earlierforstacksof arbitrarilymanyelasticplatesseparating dissimilarfluidmedia [Jackins andGaunaurd,J. Acoust.Soc.Am. 80, 1762-1776(1986)]. Thegeneralizationconsists of twoportions.The firstonemakesthelowerfluidmediumbecomeanelastichalf-space, whichmakesthemodelusefulto predict reflectionsfrom a stratifiedoceanbottom.This portionwasbrieflyanalyzedearlier[J. Acoust.Soc.Am. Suppl.180, S116(1986) ]. The second portion,nowmarriedto thefirst,consists in a second-order approxima- tionfor theparameters involvedin thedescription of thebottomreflection/transmissioncoefficients,which retainssecond-ordertermscontain- ingsecond derivatives of thepertinentparameters. Thisisanapplication to flat, layeredmediaof an (second-order) acoustical counterpartof an R-matrixformulation formany-body scattering. It isshownhowthepresentgeneralization contains theearlierfirst-order(RST) prediction asa particularcase,andhow it improvesits accuracywith a relativelyminor additional calculational effort. 2:51 DDD20. A scattering function for histuric reverberation calculations. Dale D. Ellis (Defence ResearchEstablishmentAtlantic, P.O. Box 1012, Dartmouth, Nova Scotia B2Y 3Z7, Canada) and D. R. Haller (Defence Research EstablishmentPacific, FMO, Victoria, British Columbia V0S lB0, Canada) The oceanbottomscattering functionS(O,•;O',•') depends, in general,onthegrazingangles8, 8' andazimuthalangles•b,•' of theincident and scatteredenergy.However,mostmeasurements are for backscatter only:B(8) ---S(0,•b;0Ab + •r). The few availablegeneralmeasurements indicate strong forward scatteringnear the specularangle (•'= •, •' = •), andweaker,azimuthallyisotropic,diffusescatteringawayfrom the specular.The proposedfunctioncombinesLambert'slaw with Brek- hovskikh andLysanov's surface scattering function,giving S(O,•;O',•') =/• sin0 sin0' + v( 1+ Aft)2 Xexp[ - (l/2o•)Af•], where Af• = [cos:0 + cos 20' -- 2 cos• cos0' cos($-- ½Y)]/(sin0 + sin0') 2.Byfitting/•,*,,crtothebackscattering measurements, a general scatteringfunctionis obtained.This is an improvement over two commonmethodsof extrapolatingbackscattering to generalscattering. In particular,one commonmethodusesthe separableapproximation x/B(0)B(0'), whichpeaks at •' = •r/2,ratherthanatthespecular angle. Another common method uses the half-angle approximation B[ «( 0 + 0') ] whichpeaksat thecorrectin-plane grazingangle(if 0' is replaced by •- - 0' for forwardscattering), butfailsoutof plane. S124 J. Acoust.Soc.Am.Suppl.1, Vol.82, Fall1987 arediscussed. Thescattering characteristics ofsuspended sandin a laboratorytank wereusedto providea methodof calibrationfor the ACM. Thedependence ofthescattering coefficients uponsediment sizewasalso investigated. TheACMwasdeployed inthesurfzoneatStarthope Lane, PrinceEdwardIsland,Canada, in October1984.In thesurfzone,the limitations of suchprofilometers are (a) air bubbles injected intothe waterbybreaking waves, and(b) thepossible variations in sandsizewith height above theseabed. [Worksupported byNavalFacilities Engineering Command.] 2:59 DDD22. Modeled effectsof seaflooranisotropyusingthe facet-ensemble methodand comparisonwith Red Sea data. Henry F. Schreiner(Naval OceanResearchandDevelopmentActivity, Code245,NSTL, MS 395295OO4) Bathymetricmeasurements, aswellasa knowledge of the formation processes involvedin shapingthe oceanbottom,haveindicateda strong anisotropy in bottomroughness overmuchof theworld'soceans[C. G. Fox and D. E. Hayes,Rev. Geophys.23, 1-48 (1985) ]. This anisotropy canbeapproximated bya ridgeor troughora series &ridgesandtroughs. The facet-ensemblemethod computesthe three-dimensionalcomplex acoustic pressure froma signalthathasbeenscattered andreflectedfrom thiskind of surface[ W. A. Kinney et al., J. Acoust.Soc.Am. 75, 183-194 ( 1983) ]. The modelhasbeenadaptedto a side-scanning sonarof frequency 3.5kHz andhorizontalandverticalbeamwidthsof 10'and30ø,respectively.The computation isdrivenbya 100-mssourceandsampledat a rate of 128/s.Comparisons are madewith experimentaldatafrom the highly lineat•',.d Red SeaChannel.A one-to-onecorrespondence betweenthefeatureson theechotrackandthemodeledresultsisobserved.[Work supportedby Officeof NavalTechnology. ] 3:03 DDD23.Double scattering fromtwoparallel 'rough surfaces using the facet-ensembleapproach.Henry F. Schreiner(Naval Ocean Research and DevelopmentActivity, Code245, NSTL, MS 39529-5004) Many featuresof the oceansurfaceand oceanbottomare generally long crestedin shapeand can be approximatedby thin stripsor facets placededgeto edge.Frequencyand time-domain(impulse) solutionsfor diffractionand reflectionoff this kind of surfacehave beensuccessfully modeled[W. A. KinneyandJ. G. Zornig,J. Acoust.Soc.Am. 77, 14031408(1985) ]. The introductionof a secondsurfaceof interactioncreates a significantly morecomplexproblem.Medwinand Childspresented a Huygenswaveletinterpretationfor doublediffractionoff a pair of rigid wedgesor troughs[ H. Medwinet al., J. Acoust.Soc.Am. 72, 1005-1013 (1982) ]. This interpretation is usedto constructa modelthatcomputes thecomplexacousticpressure afterscattering off two parallelfacet-construtted surfaces.The effectson the correlationbetweentwo separated receiversare presented as functionsof surfaceroughness and bottomtopographyin the frequencybandof 500-1000Hz. Also,comparisons are givenat differentseastatesof single-receiver responses from backscatter off prominentbottom featuresfor a 3-kHz 100-mssignal.All results quantifythe effectsof the roughness featureson signaldegradation. [Work supportedby Officeof Naval Technology. ] 114thMeeting:Acoustical Societyof America S124