SPECIAL FEATURE: SOUND PHYSICS www.iop.org/journals/physed Experimenting with musical intervals Michael C LoPresto Henry Ford Community College, Dearborn, Michigan, MI 48128, USA E-mail: lopresto@hfcc.net Abstract When two tuning forks of different frequency are sounded simultaneously the result is a complex wave with a repetition frequency that is the fundamental of the harmonic series to which both frequencies belong. The ear perceives this ‘musical interval’ as a single musical pitch with a sound quality produced by the harmonic spectrum responsible for the waveform. This waveform can be captured and displayed with data collection hardware and software. The fundamental frequency can then be calculated and compared with what would be expected from the frequencies of the tuning forks. Also, graphing software can be used to determine equations for the waveforms and predict their shapes. This experiment could be used in an introductory physics or musical acoustics course as a practical lesson in superposition of waves, basic Fourier series and the relationship between some of the ear’s subjective perceptions of sound and the physical properties of the waves that cause them. (Some figures in this article are in colour only in the electronic version) Introduction The main physical cause of the human ear’s perception of musical pitch is the frequency of the sound wave. Pitch is a subjective sensation and, although other factors do affect it, frequency provides its main physical or objective basis [1]. Relationships between pitches are known as musical intervals. These intervals are found within a harmonic series that consists of a fundamental and overtones that are integral multiples. The ratio between the frequencies of the first two harmonics is 2/1, this is known as a musical octave; the ratio between the second and third frequencies is 3/2, a musical fifth, then 4/3 makes a musical fourth, 5/4 a third and so on. Note that the interval names have nothing to do with the frequency ratios. They come from the number of steps of the musical scale within the interval. It is necessary to make this clear to those without musical background. See 0031-9120/03/040309+07$30.00 Table 1. Frequency ratios of musical intervals produced with a standard set of tuning forks. Interval Musical notes Frequency ratio Unison Second Third Fourth Fifth Sixth Seventh Octave C–C C–D C–E C–F C–G C–A C–B C–C′ 1/1 9/8 5/4 4/3 3/2 5/3 15/8 2/1 table 1 for a more complete list of musical intervals and their frequency ratios1 . 1 The frequencies of a standard set of laboratory tuning forks are based on the Just Diatonic Scale. There are several other types of musical scales including the Equal Tempered Scale, which is currently used in Western Music. See [1, p 171]. © 2003 IOP Publishing Ltd PHYSICS EDUCATION 38 (4) 309 M C LoPresto What follows is a simple experiment on musical intervals using a set of standard laboratory tuning forks and microcomputer hardware and software for an introductory physics or musical acoustics course. It introduces superposition of waves and the basics of Fourier synthesis as well as demonstrating some of the physical causes of the ear’s perception of the properties of sound waves. Fundamental tracking When the human ear hears a musical interval, the combination of the pitches is perceived as one pitch with a frequency equal to the fundamental of the harmonic series to which both belong, whether or not that fundamental is present [2]! Since this pitch does not correspond to any of the frequencies actually sounding, it is often called a ‘virtual pitch’ [1, p 114]. For instance, if pitches with frequencies of 384 Hz and 256 Hz (384/256 = 3/2, a musical interval of a fifth) are sounded simultaneously, the ear hears a virtual pitch of frequency 384/3 = 256/2 = 128 Hz, the missing fundamental. This ability of the ear to hear a virtual pitch from a complex tone is known as fundamental tracking. If it were not for this ability, the ear could not hear a distinctive pitch from the complex tone produced by a voice or a musical instrument [2, pp 45–7]. The sound will have a ‘quality’ that is more interesting than the dull sensation of a pure tone. This comes from the complex waveform resulting from the superposition of the two waves. The main physical cause of the perception of the quality or, as musicians call it, timbre of a sound is the sound’s harmonic structure or spectrum, which is responsible for the form of the wave [1]. When the ear hears two musical instruments or voices, even when sounding the same pitch and loudness, it can still perceive them as ‘sounding’ different. The waveforms of musical intervals can be captured and displayed using a microphone, computer interface and a data collection program2 . This is done by striking two tuning forks and allowing them to vibrate simultaneously while being held in front of the microphone (figure 1). Figure 2 shows the waveform of a musical fifth produced by two tuning forks of the above 2 Logger Pro and the Lab-Pro Interface by Vernier Software (www.vernier.com) were used. Any comparable software and hardware could be used, such as Data Studio and Waveport with the Science Workshop 750 Interface by Pasco (www.pasco.com). 310 PHYSICS EDUCATION Figure 1. Capturing the waveforms. Figure 2. The captured waveform of a musical fifth (top) and a plot of the equation y = cos 3x + cos 2x. frequencies. Using the time axis of the display to measure the repetition period of the complex waveform and then reciprocating it will give the frequency of the combination tone. The calculated frequency of the wave in figure 2 is f = 127.4 Hz, very close to the expected 128 Hz, the fundamental of the harmonic series of which July 2003 Experimenting with musical intervals Table 2. Comparison between expected ‘missing’ fundamental frequencies of musical intervals produced by tuning forks and those calculated from their waveforms. Interval Second Minor third Third Fourth Fifth Sixth Seventh Octave Frequency ratio Fundamental frequency (Hz) Calculated fundamental (Hz) 288/256 = 9/8 384/320 = 6/5 320/256 = 5/4 341.3/256 = 4/3 384/256 = 3/2 426.6/256 = 5/3 480/256 = 15/8 512/256 = 2/1 288/9 = 256/8 = 32 384/6 = 320/5 = 64 320/5 = 256/4 = 64 341.3/4 = 256/3 = 85.3 384/3 = 256/2 = 128 426.6/5 = 256/3 = 85.3 480/15 = 256/8 = 32 512/2 = 256/1 = 256 32.4 63.7 64.6 84.1 127.4 84.96 31.9 259.7 256 Hz and 384 Hz are the second and third harmonics. Using a standard set of laboratory tuning forks these complex tones can be captured and ‘missing’ fundamental frequencies can be determined and compared with what is expected from the frequency ratio of the tuning forks. See table 2 for a comparison between expected and measured fundamentals for each interval tested. Waveforms Individual tuning forks produce simple sine or cosine waves. This can easily be verified with the data collection programs and is the reason for their ‘dull’ sound quality. The expected shape of the waveform of an interval can be predicted by plotting the function that should correspond to the interval. For instance, y = cos 3x + cos 2x, is the equation for a fifth; see figure 2. The plots can be made very easily with graphing-calculator software3 by using the numbers in the ratio as the coefficient of the trigonometric function’s argument. This amounts to a simple hands-on introduction to superposition of waves or even the basics of Fourier synthesis4 . Figure 3 shows the expected waveform compared with the captured one for the interval of 512/256 = 2/1, a musical octave. Amplitude and phase differences Although the captured waveforms have close to the expected fundamental frequencies (see table 2), they may look different than expected. This is 3 Waveforms were prepared with Graphing-Calculator by Pacific Tech. (www.pacificT.com). Data Studio by Pasco has a graphing calculator as well. 4 If graphing software is not available then [3] shows the waveforms of the common musical intervals. July 2003 Figure 3. The captured waveform of a musical octave (top) and a plot of the equation y = cos x + sin 2x. because of differences in the amplitude and phase of the waves produced by the individual tuning forks. The loudness of a sound is mostly controlled by the wave’s intensity, which is proportional to the amplitude of the pressure variations that cause the vibrations of the eardrum [1]. Although every effort can be made to strike each tuning fork equally hard, producing the same loudness, small differences in amplitude that will affect the form of the combination tone are unavoidable. This is also true for differences in the phase of the individual waves as they reach the microphone. Phase differences can result in slightly different waveforms of the combination tone [4]. For instance, since the sine and cosine functions are 90◦ out of phase, the function y = sin x + sin 2x has a different waveform than the same expression PHYSICS EDUCATION 311 M C LoPresto Figure 4. The Fourier Synthesizer from NTNU Virtual Physics Laboratory. Reproduced with permission of the webmaster. with cosine functions. The forms generated with sine functions or combinations of sines and cosines may resemble those of the captured intervals more closely than those generated from cosines. This was the case with the interval of an octave in figure 3. The combination that most closely resembled the captured waveform was y = cos x + sin 2x. The plots of different combinations of sines and cosines can be checked to see which best resembles the captured waveform. It is generally accepted that although they are not totally imperceptible, unless the phase differences cause a drastically different waveform, they are usually difficult for the ear to detect [1, p 160] (see also [3, p 189] or [4, p 124]). This is the reason why the harmonic structure of a tone is considered the reason for the timbre rather than the waveform; slightly differing waveforms, which sound the same, can be produced by the same harmonic spectrum [5]. The NTNU Virtual Physics Laboratory (from www.phy.ntnu.edu.tw/java/sound/sound.html, see figure 4) has a Fourier synthesizer, which displays waveforms as the user combines sine and cosine waves of different harmonics and amplitudes while listening to the resulting sound. Adjusting the amplitudes until a waveform matches the one produced by the tuning forks provides an 312 PHYSICS EDUCATION even better way to determine the equations of the captured waveforms. This tool can also be used to test whether or not the ear can detect variations in waveforms caused by phase differences. For instance, various combinations of sines and cosines of second and third harmonics (an interval of a fifth), although varying the waveform, did not seem to noticeably change the perceived quality of the sound. Although graphing-calculator software is easiest for producing expected waveforms for comparison, a cathode-ray oscilloscope with input from a variable frequency function-generator or an actual Fourier synthesizer could also be used. Either of the latter, with speakers, could be used to perform the above test on whether phase differences perceptibly affect sound quality. Consonance and dissonance Since the time of Pythagoras, intervals that consist of pitches that have small whole-number frequency ratios have been considered, when sounded simultaneously, to be more ‘pleasing’ to the ear or, in musical terms, consonant. If the ratios are between larger numbers, the interval is perceived as less ‘pleasing’ or dissonant [1, p 55]. Although there have been many attempts to quantify consonance [1, p 156; 2, pp 167–8; July 2003 Experimenting with musical intervals Figure 5. Left: The expected waveform of a musical sixth (bottom), an imperfect consonance, compared with the captured waveform (top). Right: The expected waveform of a musical second (bottom), a dissonant interval, compared with the captured waveform (top). Note that the more dissonant waveforms are more complex. The intervals of a fifth and an octave, shown in figures 2 and 3, are consonances. 5, pp 91–3, 299–307; 6] it can still be somewhat subjective and could depend on musical training, i.e. what the ear is ‘used’ too [2, p 78]. The intervals discussed thus far, the octave and fifth, are known as perfect consonances. As can be seen in figures 2 and 3, their waveforms are relatively simple. This is also true of the musical fourth, a ratio of 4/3. The intervals of a third (5/4), a minor third (6/5) and a sixth (5/3) are called imperfect consonances; their forms are a bit more complex. Intervals of a second (9/8) and a seventh (15/8) are even more complex and are considered dissonant. See figure 5 for examples. The more dissonant intervals have more complex waveforms [3, p 193; 7], as if they are more difficult for the ear to ‘sort out’. A-426.6 Hz tuning fork simultaneously with an A-440 Hz. The frequency of the perceived pitch, the ‘fine structure’ [2, p 48] of the waveform, is 436.2 Hz, compared with an expected 433.3 Hz, which is the average of the frequencies of the two tuning forks. The beat frequency of the envelope is 13.8 Hz compared with the expected 13.4 Hz, the difference between the frequencies of the two tuning forks5 . The reason why there are two different frequency tuning forks labelled as the musical note ‘A’ is because of the existence of different musical scales [1; 2, p 17; 5, p 50; 6, p 243]. A-426.6 is ‘A’ in the Just Diatonic Scale based on C-256, meaning that it is a musical sixth 53 × 256 = 426.6 above C-256. A-440 is the pitch standard of the equal tempered scale6 . Beats Note that the waveforms of the less consonant intervals in figure 5 begin to resemble the familiar pattern of beats. This is to be expected, as beats are considered very dissonant. When musicians ‘tune’ instruments they are attempting to play at the same frequency as one another. Beats are what they are listening for when they attempt to determine whether or not they are in tune. Figure 6 shows beats produced by sounding an July 2003 Dyads and triads When two musical pitches are sounded simultaneously the resulting tone is called a dyad; when three are sounded it is called a triad. The intervals 5 Reference [8], an accompanying laboratory manual to Vernier Software’s equipment, has an experiment on beats on p 21-1. 6 Reference [8] also has an experiment on the frequencies of the equal tempered scale, Mathematics of Music, on p 23-1. PHYSICS EDUCATION 313 M C LoPresto Figure 6. The pattern of beats between tuning forks of frequencies 440 Hz and 426.6 Hz. Table 3. Comparison between expected ‘missing’ fundamental frequencies of triads produced by tuning forks and those calculated from their waveforms. Triad Musical notes Frequency ratio Fundamental frequency (Hz) Calculated fundamental (Hz) Major Minor C–E–G E–G–B 256/320/384 = 4/5/6 320/384/480 = 10/12/15 256/4 = 320/5 = 384/6 = 64 320/10 = 384/12 = 480/15 = 32 63.7 31.9 that have been discussed thus far are all dyads. When three pitches of frequency ratio 4/5/6 are sounded together this is known as a major triad; the ratio of 10/12/15 is a minor triad. These two common types of ‘chords’ are the basis for much of the harmonic structure of Western music. Since a third tuning fork was involved, it was no surprise that it was harder to capture the expected complex forms of triads than dyads, but the expected fundamental frequencies could, however, still be calculated with just as much accuracy as the dyads. Figure 7 shows the expected and actual shapes of a major triad of frequencies, 384/320/256, fundamental 384/6 = 320/5 = 256/4 = 64 Hz. The measured fundamentals in table 3 are very close to what was expected7 . Conclusion The exercise of capturing the combined waves, observing their forms, calculating the fundamental frequencies and making comparisons with the expected waveforms and frequencies can provide a hands-on and practical lesson in superposition of waves and even an introduction to the 7 Reference [3, p 195] also has waveforms for triads if graphing software is not available. 314 PHYSICS EDUCATION Figure 7. The expected (bottom) and captured (top) waveforms of a major triad. basics of Fourier composition. Inspecting the effects of amplitude and phase and exploring consonance and dissonance can help to provide understanding of the relationships between the subjective properties of sound perceived by the ear and the objective ones produced by the waves. July 2003 Experimenting with musical intervals Note. Five Logger-Pro files that were used in the experiment are available from the electronic journal (stacks.iop.org/0031-9120/38/309). Acknowledgments I thank HFCC honor’s student Leisanne Smedala for asking the questions that originated this project and Honor’s Program Director Nabeel Abraham for allowing her to work with me an extra semester. I also thank Dr Leroy S Williams, retired Professor of Music from Edinboro University of PA, and HFCC music instructors Kevin Dewey and Randy Knight for providing essential perspective from the musical side of things. [3] White H E and White D H 1980 Physics and Music: The Science of Musical Sound (Philadelphia, PA: Saunders College Publishing) p 192 [4] French A P 1971 Vibrations and Waves (New York: Norton) p 25 [5] Sethares W A 1999 Tuning, Timbre, Spectrum, Scale (New York: Springer) p 16 [6] Rigden J S 1985 Physics and the Sound of Music 2nd edn (New York: John Wiley) p 239 [7] Helmholtz H 1954 On Sensations of Tone (New York: Dover) p 196 [8] Vernier et al 2000 Physics with Computers (Beaverton, OR: Vernier Software and Technologies) Received 14 February 2003, in final form 12 May 2003 PII: S0031-9120(03)59582-X References [1] Rossing T D 1990 The Science of Sound 2nd edn (New York: Addison-Wesley) p 80 [2] Roederer J G 1995 The Physics and Psychophysics of Music—An Introduction 3rd edn (New York: Springer) p 44 July 2003 Michael C LoPresto is chair of the Physics Department at Henry Ford Community College in Dearborn, Michigan and is an amateur trombone player and vocalist. He is also the author of an astronomy laboratory manual and a textbook supplement ‘Cycles in the Sky’. PHYSICS EDUCATION 315