'sTOF TECHN 01
AN
AUG 3 1965
1905
3·PF
*O
IMPROVED SOLID STATE TERMINAL
ANALOG SPEECH SYNTHESIZER
by
,yAYMO,.) SAMUEL TOMLINSON
B.SE.E.
Rensselaer Polytechnic
(1963)
Insti tut
SURMI TTED IN PARTIAL FULFILLMENT OF THE
YEQlU IREME\TS
Tir
COR
nUo
',MASTFR OF SC IFNCF
at
.ASSACHUSETTS
Si-nazt.u•re
the
INSTITUTF OF TECHNOLO'Y
June, 1965
of Auth or
nepartment of
Certified by
Elect
rical
En inecr!n,
- f '
9
'i~
2 ,i
o r"-
Thesis
Acceptp
d by
,-- Cha rmrn, Departmental
---
--
Surnrvisir
R
Committee ome Graduate Stu-e'n-
38
M
AN IMPROVF0 SOLID STATE TERMINAL
ANALOG SPFECH SYNTHESIZER
by
RAYMOND SAMUEL TOMLINSON
Submitted to the Department of Electrical Engineering on
May 28, 1965 in partial fulfillment of the requirements
for the degree of Master of Science.
ABSTRACT
This report discusses the application of modern technology
to developing an improved terminal analog or resonance
speech synthesizer. The concept of hybrid computation is
used to combine the speed of analog circuitry with the
precision and flexibility of digital control.The familiar
pole analog is reali ed with anai~nr crroute technscad;;
niq.rs
i;
which the circ•tit
parametprs
larp controlled with
The .xperimenter
rigital signals from a small computer.
exercises control over the synthesizer through on line commands typed in at a typewriter console. The ability to produce an utterance and almost .inst:antly jud.ge i:ts quality,
modify it, and try again, allows the d.esired restult to he
attained much more -rapidly than has been previously possiblo.
flexible data input is provided through several sources,
including a light pen with which the required control sinal
curves may be drawn on a CRT display.
Provision is made for
processing the input data in a general way to allow the
specification of something other than formant frequency or
glottal frequency.
Thesis Supervisor: Kenneth N. Stevens
Title:
Professor of Electrical Enginpeering
S-i-
ACKNOWLFEDGEMFNT
The author wishes to exp.ress his thanks to
Professor Stevens for his aid, opinions and
guidance. Thanks are also extended to the
members of the Speech Grou p for their assistance
Particular thanks are
in preparing this thesis.
due Mr. Keith North for hi s invaluable help in
assembling the hardware, M r. James Williams for
designing and constructing some of the circuits
we have used, and Mrs. Jan e Heinz, Miss Eleanor
River and Mr. Nod Renhaim for their aid in
This work would not
preparing the manuscript.
have been possible without the support and
facilities of the Speech C'
ommunications Group,
the Research Laboratory of Electronics and their
supporting agencies.
-
ii-
TABLE OF CONTENTS
Chapter One - Outline of the Problem
1.1
Introduction ................................
1
1.2
What Makes a Speech Synthesizer .............
1.3
Hybrid Computation .........................
3
4
1A4
Programming Considerations ......
7
7......
Chapter Two - Hardware Design
10
2.1
Overall Synthesizer Description ............
............. 11
Pbrmant Circuits ............
2.2
2.3
The Pole-Zero Pair Circuit ..................
12
2.,
2.5
2.6
2.7
Perturbations Lhe to Parasitic Phase Shift ..
Glottal Source ..............................
Noise Generator and Modulator...............
Digital Attenuators .........................
16
20
21
22
2.8
Interconnecting SWirtches
.............
. 22
Control Logic ...............................
Miscellaneous Circuits ......................
2.10.1
Overload Detectors .................
Higher Pole Correction .............
2.10.2
2.10.93
Output Amplifier ...................
Chapter Three - The Synthesizer Control Program
Overall Program Description .................
3.1
2.9
2.10
3.2
28
Description of Part One .....................
28
Commands and Data Storage .........
3.2.2 Data Input ...
..................
Data Display ........................
3.2.3
3.2.4
Calibration of the Synthesizer ......
3.2.5
Other Features ......................
3.2.6
Error Comments ........ .............
Operation of the Editor ..............
...
.
Operation of Part Two ......................
3.4.1
Expression Evaluation ...............
29
31
34
35
35
35
36
37
39
3.2.1
3.3
3.4
23
25
25
26
27
3.4.2
3.4.3
3.5
Special Events Compiler .............
Error Comments ......................
40
41
Operation of Part Three ..................... 42
Chapter Four - Performance Evaluation
4.1
The Scope of the Evaluation .................
4.2
Technical Performance ...........
4.2.1
4.2.2
4.2.3
4.2.4
4.3
References
45
........
46
Pole-Zero Pair Circuit ..............
Formant Circuits ....................
46
Interconnecting Switches ............
The Overall Synthesizer .............
48
Listening Evaluation ........................
48
49
50
LIST OF PIGURES
2.1
Block Diagram of Complete Synthesizer.
2.2a
Analog Computer Simulation of Resonator.
2.2b
Realization of the Resonator Using Operational
Amplifiers.
2.3
(does not exist)
2.4
First Realization of Pole-Zero Pair.
2.5
Modification of Figure 2.4 to Allow Pole Frequency
to be Less Than the Zero Frequency.
2.6
Block Diagram of Final Pole-Zero Circuit.
2.7
Detailed Circuit for Pole;Zero Pair.
2.8
Simple Resonator Diagram.
2.9
Robt Locus Diagram for Simple Resonator.
2.10
Differential Integrator.
2.11
Pole Zero Plot for Non-Tdeal Differential
Integrator.
2.12
Phase Angle of Non-Ideal Integrator.
2.13
Root Locus Diagram for Resonabor with Non-Ideal
Integrabbr (no,.,damping),
2.14
DeterminAtion of Phase Sb~tft due to Extraneous
Pole and Zero.
2.15
Attenuator Stage Operation
2.16
Interconnecting Switch Circuit.
2.17
Overload Circuit Sdhematlcs.
2.18
Output Amplifier.
3.1
(does not exist)
3.2
Pushdown Evaluation Flow Diagram.
3.3
Special Events Compiler Flow Diagram.
LIST OF FIGURES - con't
4.1
Pole Frequency Accuracy Curves.
4.2
Zero Frequency Accuracy Curves.
4.3
Bandwidth Accuracy Curves.
4.4
Synthesized Vowels and Diphthongs.
4.5
Synthesized "Where are you?"
4.6
Assorted Stop Consonants.
4.7
"Are you a good boy or a bad boy"
TABLES
4.1
D.c Evaluation Data.
4.2
Interconnecting Switch Data.
4.3
Signal to Noise Measurements.
Outline of the Problem
Chapter One
1.1 Introduction
Before
going into the "how" of speech synthesis, a few
speech
in
is an end
the
synthesis
Second, speech
testing
theories of
receiving
is an
the
human speech
from the
using
is capable of
means
important
accurate
Speech synthesi-
for improving the understanding
production process.
of
A theory of
guiding an
synthesis is better than one which cannot.
zers are valuable tools
end
of
be able to generate speech.
human speech production.
speech production which
means
some
Man-machine communication
requires the machine to
of
communication using
requires
the
at
speech
transmitted signals (1,3).
speech
Speech
itself.
techniques
bandwidth compression
regenerating
First, the generation
in order.
words on the "why" are
of
Speech synthesizers
are also useful in studies of speech perception because they
can generate well defined signals and manipulate its parameters to examine human responses to stimuli so generated.
Of the
have
been
of speech
different types
constructed,
the terminal
-I-
synthesizers
analog
which
or resonance
synthesizer is by
reason
for
this
far
popularity is that
synthesizer is relatively
synthesize
most
easy
speech
The
popular.
the most
the
to use,
sounds.
synthesizer is considerably less
principal
terminal analog
and
is able
Furthermore,
complex than
such
most
to
a
other
types.
As
more sophisticated
and precise
proposed, existing synthesizers have
precision,
accuracy,
range,
and
theories have been
been found lacking
stability.
in
This thesis
reports the development of a new solid-state terminal analog
speech
synthesizer with marked
In addition, a great deal
improvement in these areas.
of flexibility has been
achieved
through the use of on-line computer control.
It is not our
intention that this
treatise on the general problem
thesis be either a
of speech synthesis, nor
a
detailed description of the many operational features of the
synthesizer.
Rather, we
hope to present
the
capability
of the
general
detailed look at
Separate
a description
synthesizer,
certain important aspects
plus
of the
of
a more
design.
documents are being prepared to serve as operation
manuals for those who wish to use the synthesizer.
-2-
1.2 What Makes a Speech Synthesizer
The basis of all terminal analog speech synthesizers is
that
the spectrum of the
signal can
be separated
into a
source function and the transfer function of the vocal tract
(10,±1).
Furthermore,
Fant
(11) has
acoustic tube driven at one end
transfer
for an
and open at the other,
the
function can be approximated as closely as desired
by a cascade of
pole
shown that
pairs.
circuits containing only complex
That
such a model
conjugate
works is proven
by the
existence of many working speech synthesizers which use this
principle.
A model consisting solely of cascaded complex conjugate
pole pairs is only adequate
The
two notable examples
for certain classes of
where this model
sounds.
breaks down are
sounds in which the tube is excited other than at the
and
when more
than
one acoustic
tube
end,
is involved. An
analysis of such acoustic configurations (16,23,24) produces
a transfer function which contains
pairs as well as pole pairs.
sounds
complex conjugate zero
If a synthesizer is to produce
with spectra containing
zeroes, a transfer function
containing zeroes must be included.
There are
speecht
two
principal excitation
a quasi-periodic pulse
-3-
sources
in most
source caused by vibration
of the vocal cords (glottal source), and random noise caused
by turbulence as the airstream passes through a constriction
in the vocal tract (noise source).
synthesizer
contain
should
of
kinds
both
should
source
noise
addition, the
A general purpose speech
be
sources.
In
by
the
modulated
source (8) since, in the vocal tract, the intensity
glottal
varies with
of the
noise
which,
in turn,
cords.
Other sources
is varied by
of the vocal
the vibrations
to produce pops
airstream
the
velocity of
the
and clicks could
be
included, but this has not been done in this synthesizer.
In
addition
the above,
to
there
several other
are
auxiliary features which might be included.
These
features
are not essential to the synthesis of speech, but contribute
to adjustment, testing,
Several
such features
or monitoring
are included in
of the
synthesizer°
the present synthe-
sizer, and will be discussed later.
1.3 Hybrid Computation
The motivation
(5,6,22)
behind the
that analog
is
use of
computation
hybrid
computation
is superior
(at the
present time) to digital computation for certain tasks,
vice-versa.
and
In particular, analog computation is especially
suited to high
speed linear processing
-4-
of time
functions.
The ease with which most filtering problems may be solved by
bringing together the
right assortment
cuits is fairly well established.
conceptually
speed computers presently
even the highest
also, but
cir-
The solution to this same
digital computers is
problem using high speed
easy
of electrical
available cannot perform arithmetic at a rate fast enough to
adequately process these time functions in real time.
When
non-linear
tempted, the picture
analog
circuitry
or
time-variant
starts to change.
instantly increases.
computation
is at-
The complexity
Good,
of
high speed,
wide range multipliers and similar circuitry is prohibitively
expensive.
The complexity
creases somewhat,
but there
of digital
computation in-
is no additional
cost.
problem of computation speed still exists however.
The
When the
problem of communicating with the user is considered, analog
computation
is an
extremely
weak
communication cannot
be
About
can be done
the best that
"sign language"
in
achieved with
which
pictures to each other.
contender.
analog
is by means
the user
and
Effective
techniques.
of a kind of
synthesizer
draw
Such a scheme cannot compare to the
tremendous flexibility offered by a digital computer capable
of
handling both "sign languaget
symbols which can
be combined into
communication, as well as
higher level
languages
including languages composed of a subset of English.
- 5*-
The obvious next step is to combine the better features
of both digital and analog computation.
is referred to as hybrid computation.
cation with the user
along with
can then be
Such a
combination
The task of communi-
assigned to the
computer
certain information processing tasks,
and the
signal processing performed by the analog circuitry.
In the case of speech synthesis, the signal
is linear and time-variant.
be controlled by
described
the
processing
The time variation is what must
digital
computer.
a method of controlling
Henke
a terminal analog speech
synthesizer through a digital computer. Because the
circuits
he
used were
previously
control voltages, he achieved
(4) has
built to
analog
accept analog
control over the
synthesizer
with analog signals from a digital to analog converter.
If the intention
is to
control the
synthesizer from
digital signals, a more direct approach is possible.
scheme has been described by Whitman (5).
Such a
The parameters of
the analog circuit are set with "coefficient multipliers" at
particular
points
multipliers
in the
simply
signal path.
multiply
(amplify or
signal by a particular amount.
synthesizer, the
discrete
amount
of
These coefficient
In
attenuate)
the
the method used in this
multiplication
is varied
in
steps by the use of transistor switches controlled
by digital signals.
-6
the intermediate
In bypassing
one
major deterent
to high
quality synthesis
in its
circuit is eliminated, and
analog
is removed.
in a
speed analog multiplier necessary
The high
signal,
analog control
strictly
place a digital
built
multiplier can be used. Such a multiplier (7) can be
to perform with high precision at very moderate cost.
1.4 Programming Considerations
currently exist two
There
Closed shop operation in which the user's program is
usage.
run
by a machine operator and the results returned later is
characterized by
results
(the delay
turn-around times
received) ranging
are
one or
from
before
two
program is characterized by
the
hours to
On-line operation in which the user runs
several days.
own
prevalent modes of computer
his
a fast turn-around time of
minutes or seconds at the expense of higher computer usage.
For
an application such as
the speech synthesis being
discussed, the advantages of on-line operation outweigh
The availability of
disadvantages.
makes
such
computer
in
operation
the
even
more
small computers at MIT
attractive.
Research Laboratory
chosen for four reasons.
of
The
-7-
PDP-i
Electronics was
First, a cathode-ray tube
is available to present pertinent
any
display
data in a meaningful and
easily interpreted way.
may
be used
for storing large
special panel
user's
has been
external
Fourth, it
drum
Second, a high speed magnetic
amounts of
added to
1-0 equipment
in
Third, a
to
the computer
easy to
is relatively
data.
an almost
connect
trivial way.
time
schedule
on
the
computer.
functions of
The
synthesizer may
program
be
the control
into
grouped
the speech
program for
three
The
categories.
must accept input data or assist in its production,
provide information to
when he
the user
requests it, and
control signals for the. synthesizer to produce an
generate
utterance.
To maintain generality, it is desirable to divorce
operation
of the program,
as it appears
the actual synthesizer as much
be
having to
rewrite
the
to the user, from
as possible.
possible to make alterations
program.
the
And it
should
to the synthesizer without
The
utilizes
program
a
"mapping specification" to govern the relation between input
and output data.
in
advance to
This
specification is typically
be read
in by the
user.
By
prepared
doing so, the
pecularities of the synthesizer are taken into consideration
without
the user having to worry about it.
the user wishes to change
may
In addition, if
the appearance of the machine
do so by altering the mapping specification.
-8-
he
More will
be said about this feature in chapters three and four.
with
for
necessary
ability,
or even to
such basic things
this
end,
user
a potential
know how to
self-contained.
The
to
not be
operate a computer beyond
switch is.
has been written to
entire
usable,
programming
possess
what a sense
as knowing
the program
it should
by anyone,
of training,
a minimum
program to be
is for the
Since the intention
operation of
To
be completely
the
program
is
There is no necessity
controlled through typed in commands.
for manually starting the program or using any other program
in
any
way. The
to
program tries
recover from
its own
mistakes or mistakes made by the computer.
The command
language
has
been made
possible to allow rapid type-in of commands.
language is not quite
as
mnemonic as
as
concise
as
Such a concise
one which
is less
concise, but after using the program for a while, most users
have no difficulty remembering which command does what.
command
language
is discussed
three.
-9-
in
more detail
The
in chapter
Chapter Two
Hardware Design
2.1 Overall Synthesizer Description
The synthesizer hardware can be divided into components
having specific functions such as the noise generator or the
pole-zero
pair circuit.
Mechanically, each of these compo-
nents is entirely contained on
cards.
The sections
one or more plug-in
which follow
various components in
detail.
describe each
circuit
of these
the
section describes
This
overall synthesizer.
A
block diagram of the analog circuitry is illustrated
in figure
2.1.
There
are
several possible
signal
paths
through the synthesizer according to the type of sound to be
produced.
The
path to
transistor
switches.
nants which
can be
consisting
be used
For
is selected
vowels, glides,
characterized
by a
of complex conjugate pairs
by a
set
and some conso-
transfer
function
of poles (CCP poles)
driven by a pulse source at the glottis, the signal path
indicated
with solid lines.
of
is
Similarly, aspiration, frica-
tion (voicea ana unvoicea), and nasalization are provided by
other
paths through the synthesizer.
on figure 2.1
according to
These paths are drawn
the legend on
-10-
the figure.
It
should
be
noted that
the output
inputs to
amplifier are
mixed (added) together to produce the output.
2.2 Formant Circuits
an
The formant circuits used were originally built for
experimental forerunner of the present synthesizer.
of the design are given in reference (6).
is shown in figure 2.2.
Details
The circuit
used
The bandwidth of the circuit is not
programmable, and must be set by a potentiometer (R').
center
frequency
is
determined
by
a
digital attenuator
When the
between the first and second integrators.
ator
The
attenu-
center frequency is G, the gain
setting is 0 db., the
As the attenuation is increased,
factor of the integrators.
the center frequency decreases.
The transfer function for the circuit is
+
A G+
Thus
The
attenuation.
the
as the square root of the
natural frequency (• varies
Since the maximum
attenuator setting is one,
maximum frequency of resonance
gain factors of the two
is G.
In practice, the
integrators are not equal, and
maximum frequency is the geometric mean of the two gains.
-k
'-
the
The
maximum frequencies
chosen for
the four formants
are 1600 cps for Fi, and 5000 cps for F2, F3, and F4.
frequencies
are
far
above those
normally
These
encountered in
speech, but the extra range might be useful for studying the
of infants,
speech
or non-speech
minimum frequencies are 41 cps for
F3,
and F4.
sounds.
Fi, and 128 cps for
at their
When operating
The theoretical
lower extremes, the
formant circuits tend to become noisy due to the lower
A practical minimum for
gain.
F2,
loop
these center frequencies is
twice the theoretical minimum.
2.3 The Pole-Zero Pair Circuit
From a consideration of the configuration of the
tract
vocal
during the production of nasals and nasalized sounds,
and from
spectrum matching experiments
(23,24),
it is evident that
only poles does
spectrum
for
these
sounds
a transfer function containing
not comprise a very accurate analog.
The
of the noise produced by turbulence at a constric-
tion in the vocal tract in the production of fricatives also
contains
zeroes (16) due to the effect of the cavity behind
the constriction.
necessary
to
For the production of these sounds, it is
introduce zeroes
produced by the synthesizer.
-12.-
into the
transfer function
been
In the synthesizer being described, a circuit has
transfer function consists
impossible
cies, an
actual
must
circuit
frequencies.
where their
effect on
a
have
frequency
where
There
transfer
are two
the signal
function
may
be
the
Two
a
The
factors
need for a CCP pole
the fact that
filter (16), and
or at
useful°
more satisfactory.
influenced this decision are the
in the noise
choices for
is negligible,
their effect
latter approach was deemed
pole
gain at high frequen-
They may be at very high frequency
location of these poles.
which
physically
poles as well to cancel the effect of the zeroes
at very high
lower
zero is
solely of a
to the high
to build due
containing
Since a circuit whose
to satisfy these needs.
incorporated
if the
extra
is not needed, it can be raised to a high frequency so
that its effect is minimized.
The circuit to be designed is
to provide a pair of complex conjugate zeroes, and a pair of
complex conjugate poles.
The first
obvious
circuit considered
method
of
building
a
feedback path
contained
one
designed
producing
a pair
for
equations are developed below.
-t 5-
of
for this
feedback
the
of
role used
amplifier
circuits
poles.
the
whose
previously
The pertinent
Referring to figure 2.4a, we
A
e,,t
zeroes of
The
.
AA=
s++a&,S + (I
ei,-n
,)
~
out
m
eaut =
write
)
poles of the
the transfer function are the
feedback function F(s), as we would expect, and the poles of
the
transfer function are those
multiplied by square
widths
of both
drawbacks are
of the
root (1+A).
the pole
Furthermore, the
and the
immediately
feedback function
zero are
apparent..
First
band-
the same.
the
Two
transfer
function goes to zero when the pole frequency is the same as
the zero
must
lie
negative.
by
frequency (A=O), and second,
above the
zero frequency unless
frequency
A can
be made
Both these problems can theoretically be overcome
judiciously
reordering the
making the gain go negative.
in
the pole
figure 2.5.
components
and effectively
The circuit for this is shown
The relevant equations for this circuit are
given below.
e,,t
=
Gein -
G (K e0
=
2_t
't
ISO's+
An experimental circuit revealed that parasitic phase
in
various parts of the
e.t
+
W
,
shift
circuit introduced wild deviations
from theory in the real part of the poles and zeroes.
it may have been possible to compensate
While
for the parasitic
phase shift, it was felt that a different approach would
be
better.
Writing out the desired transfer function as a function
of I/s, we gett
L +W
From the
diagram
s5+
b
C -+.-
+
last equation,
we can
ein
I
immediately draw
of the required circuit
(figure 2.6) consisting of
integrators, summers and coefficient multipliers.
coefficient
multipliers can provide no
supplied by the integrators.
then
be worked out
a block
gain, the
Since the
gain is
The details of the circuit may
in a standard
fashion.
The problem is
not completely solved though, since there is still parasitic
phase shift to be compensated.
The general problem of phase
shift is discussed in the next section.
The final
transfer
circuit, shown
in
figure 2.7,
provides
a
function consisting of a conjugate pair of poles,
and a conjugate pair of
zeroes.
The center frequencies
of
the pole and the zero are individually variable from 255 cps
to 10000
cps.
The
bandwidths are
from 10 cps to 200Q cps.
independently
variable
The bandwidth is constant + 20 cps
over the
entire frequency
with the
range
greatest
error
occurring above 3000 cps.
2.4 Perturbations Due to Parasitic Phase Shift
Since the operating frequencies of the circuits used in
the
synthesizer
conventional
introduced by
purposes
of
are
considerably
operational amplifier
parasitic phase
discussion,
we
resonator section in figure
tion
of the poles by the
For the circuit
the poles
move along
than
those
more acute.
refer
to
the
2.8 and consider the
simple
presence of unwanted phase shift.
2.8, the root
locus is
parallel to
a line
the j-axis.
axis.
If we now introduce a demon which produces a
hp shown
shift without disturbing
(201)
-hat
If
the locus coincides with the jcertain
the amplitude of
the loop gain, the closed loop poles will be perturbed.
nn
as
As the loop gain (A) is increased,
is no damping (T= ),
of phase
For
perturba-
there
amount
in
circuitry, the problems
shift are
shall
in figure
in figure 2.9.
pictured
higher
at the frenu•~v of any nol
It
of -- th
closed loop, the loop gain must be -1 for negative feedback.
Graphically, this implies that the angles of all lines drawn
to the
those
closed loop
pole from
the open
loop zeroes
minus
from the open loop poles must be congruent modulo 360
must
damping, the
poles must
closed loop
as the phase
at the origin
centered
For
sum as well.
be included in this
the case of no
move along
circle
a
shift produced by the
circle
along this
The distance traveled
demon is varied,
shift
present, their phase
If demons are
to 180 degrees.
must be such as to change the angle from the open loop poles
at the origin
produced
by
circular arc
by an
amount which cancels
If
the demon.
may
this distance
be approximated
by
the phase
is
a straight
shift
small, the
line
of
length P.
p
-
To get a feeli:ng for the allowable phase shift, consider
w
= 2-rr (1,0000
cpS)
P = 2T (10 Cps)
~then
2.o
It
10,000o .002 racdwns or .A die.
sources
Certain
~I- be immediatelv
- of
-- ~ phase
----- error may
-~r rrr~
I~
digital attenuators (7), operational
enumerated as followst
amplifiers (5,6), other
components, wiring.
these
produce essentially linear phase lag by introducing
sources
very high frequency poles.
introduce
lead.
All of
zeroes in
The method of compensation is to
the loop
This is accomplished
to provide
cancelling phase
by adding series resistors
in the integrating capacitors.
-IT-
(6)
If
this
were the
would be no problem.
obvious,
only source
There
of phase
shift, there
is another source which is
not
and which is the most significant source of error.
Consider the differential
integrator of
figure 2.10.
Its
transfer function is found as follows.
e-
o
Ri
e, ..
R, +
-
A,
RC
Rj+ sC,
yC2
"rZ= R2. C2.
I"+-
assume A
o
1,
eo
Gil
infinity then
-
e..
-
et,
+tSr
s
+
eo
4+TS-rj
assume el = 0
if
and
%
it
Tthen
_
is a simple integrator,
If,
-. 'r•
-
•,
then there is
an extraneous real axis pole and zero added to the
integra-
tor.
The
shown in
pole-zero
figure 2.11.
frequencies
The phase
is shown in
figure
integrator is included in a
the
integrator is
this non-ideal
plot for
contribution for
2.12.
If
this non-ideal
loop with an ideal
root locus in figure 2.13 results.
-Is8-
J-axis
integrator,
The asymptotes (20)
are at +
90 degrees and
loop poles and
open
centroid of
pass through the
The location
zeroes*
the
of the centroid
then gives the maximum perturbation of the real part of
the
closed loop poles.
A
estimate
rough
of
perturbation
the
Refer to
found as follows.
realistic situation is
for
a
more
figure
2.14.
Tj = 1I
\
I
6r=
oI
The
phase shift, on the
·t-i
.
J-axis due to 'the
pole and zero at
frequency w is
=
=
el-
- Circtan
orctan
G
orctcn (L)
- ac'rcan O -)
approximating the difference as a differential
AT
Ass
-
-p----dr L'c-ct0r4-'
T,
for
LO=1r
(re vc~L\
W
IL
,2.
u
If
2
k
nact ýr lk#tct{)
i aunty
rc6o
9
CM rn
vrp
t*VIe.
6
2-
j
2.
•= .ot1, (one percent error in time constant)
-19-
then
.005 r-adians
=
.tOr
=
The perturbation for
=
.P
I-M- t(O÷
2.
-r. 5
34
'B
ec
alo
cPS
To achieve an error of + 20 cps, requires adjusting the time
constants to be equal within + 0.4 percent.
2.5 Glottal Source
The
glottal source consists of a single shot triggered
by the computer to produce a narrow rectangular pulse.
spectrLUm
i
band.
A filter
consisting of two
poles
is
to
s
used
pulse
is.;
shape
The
shaping filter is combined
filter
and the
was made
to
.LaC.
tesseiitiaLC.lly
spectrum.
Luio.LV
these
axis
Presently, this
with the higher pole
any of
iLth
low frequency real
radiation compensation
design
VtVr.
The
filter.
filters,
correction
No attempt
and
developed for an exploratory model (6) were used.
circuits
The final
model will have these filters separated.
After being shaped, the glottal pulse is attenuated
a
digital
attenuator
to
determine
attenuated signal is connected t.o
its
amplitude.
the input of the
circuits by one of the interconnecting switches.
-20 -
by
The
formant
2.6 Noise Generator and Modulator
Our original intention was
in
the synthesizer
The
sources.
noise
as part of
however, due
been possible,
hard limit
this thesis.
in
available
for modulating the
be used
idea
Briefly, the
elsewhere (8).
noise with
wide-band
circuit
This has not
to shortcomings
scheme to
noise has been described
is to
to include a noise
a
limiting value
Low-pass filtering the
determined by the modulating signal.
result yields essentially band-limited Gaussian noise.
problem
The
signal
if the
component in
the
distribution
is asymmetrical.
may have zero
than
mean, it has
the other.
then have a d.c.
sign
of
signal
through
variations
d.c.
to
developed.
the d.c.
component
the modulating
any of
the filters
time
only on
the
expedient of passing the
not
signal.
work due
The
to
the
application of
causes unwanted
Therefore, more work
from the output.
more of the
since it depends
a capacitor will
amplitude
limiting process will
of the
The simple
in
source
is, although the noise
one sign for
The output
its input.
noise
That
a d.c.
production of
to the
arises due
noise to
be
is necessary to eliminate
2.7 Digital Attenuators
The design
quately
of the
has been ade-
digital attenuators
A brief description fol-
described elsewhere (7).
lows for completeness.
The basic principle of one
illustrated
in
figure
stage of the attenuator
By
2.15.
or
opening
closing the
switch, one of two attenuation values may be selected.
the
When
switch is open, the signal passes through unattenuated.
When the switch is closed, the attenuation is determined
the
is
voltage divider ratio.
by
stages followed by
Eight such
unity gain buffer amplifiers are cascaded to form a complete
attenuator. The attenuation provided by the eight. stages is
1/4, 1/2, 1, 2, 4, 8, 16, and 32 decibels.
attenuation
The
of the complete attenuator is the sum of the atteuations (in
decibels) of
switch
those
with the
consists of a single
off or saturated
various
stages
switch
closed.
The
transistor which is either cut
according to
a digital
level.
By using
combinations of digital levels, the attenuation may
be varied from 0 to 63 3/4 db in 1/4 db steps.
2.8 Interconnecting Switches
The interconnection of the various parts of the synthe-
-2 2.-
is done
figure 2.1
illustrated in
as
sizer
digitally operated transistor switches.
states.
by a set of
Each switch has two
In the "on" state, a connection is established from
the input
to
output.
In
the
"off t state,
there
is no
connection.
When
the switch is off,
disappear as
impedance
load
as the
far
the switch should effectively
shouldbe high in the off position.
output
The
is concerned.
This requires
a more complicated series switch rather than a shunt type.
In addition, the offset should be low (less than 5 mv.), and
a minimum of
control current
source or the load.
this.
should appear
operate
amplifier for
applied
the
to the
the
The circuit in figure 2.16 accomplishes
Its operation is briefly as follows.
transistors
in either
as
a phase
digital
splitter
signal.
digital input,
When
the switch
The first
four
and push-pull
zero
volts
is
transistors are
back-biased to 6 volts and switch is open. When -3 volts is
applied
to the
input, the
transistors are
forward biased
through the 20k resistors closing the switch.
2.9 Control Logic
The functions of
and
general control
modules.
data buffering and routing,
are provided
by an
timing,
assembly of logic
The control logic is composed of four parts.
The
these
outputs of
The
registers.
devoted to
logic is
of the
bulk
tion switches.
flip-flops are
used
to
the attenuators and configura-
the digital inputs of
drive
thirteen 8-bit flip-flop
The contents of these registers are
altered
by commands from the computer.
eight bit counter is used
An
of time.
When
The counter is wired to count down to zero.
counter is
the
as a clock to keep track
stepped below
overflow pulse is
zero, the
used to set a status flip-flop which signals thecomputer and
The counting pulses are produced
inhibits further counting.
by a 100
kc oscillator.
of the
The purpose
clock is
to
avoid the necessity of keeping track of time in the computer
program.
The signals sent to the synthesizer consist of 18
from
pulse
the IO register
generated
instruction
by
of the computer
the
execution
in the program. A set
increase, the driving capability of
computer
and to provide the
bits
plus an i-o transfer
of
a
particular
iot
of inverters is used to
the data lines from
the
inverted sense of the signals.
The pulse from the computer is also cleaned up with a
pulse
amplifier to assure reliable operation.
An address decoder is used to decode bits 1-5 of the IO
register.
at
The decoded levels are used to gate the iot pulse
the input of
a pulse amplifier.
-24-
The output pulse from
the pulse amplifier
into
the selected
are used
to set
is used to
Jam-transfer IO bits
buffer register.
the clock,
Three
10-17
of these pulses
generate glottal
pulses,
and
generate marker pulses.
Two
levels
synthesizer.
the
are sent
back
One signals that
to the
computer
the clock has reached
other signals that an overload has occurred.
levels are taken
priority
is one,
proper
as status
bits in the
interrupt system of the
the
to be
taken.
computer.
The
completely straightforward and
design
will not
zero,
These two
sequence break
or
When the level
normal program is interrupted to
action
from the
allow
of the
the
logic is
be discussed
fur-
ther.
2.10 Miscellaneous Circuits
2.10.1 Overload Detectors
Due
to
the
high
synthesizer,
the
possibility
levels
formant
exists.
Q of
the
of
tuned
developing
This is particularly true
frequencies
are
close
circuits
together.
in the
high
signal
if two
of the
After
passing
through succeeding stages of the synthesizer, the distortion
may no longer be discernible through casual listening.
i
In order
the occurrence
detect
to
overload
and
the operator, an overload detector has been incorpo-
signal
The circuit for accomplishing
rated into the design.
developed by Jim Williams
was
of
of the Speech Communications
The-circuit is shown in figure 2.17 and operates
group.
The input signal
follows.
this
is applied to
as
the bases of two
transistors operating as level discriminators.
The emitters
are biased at + 5 volts such that if the signal exceeds this
will conduct
voltage, a transistor
across
the
common load
resistor.
then sets a flip-flop whose
an
overload has
a voltage
developing
This
pulse-like signal
output serves to indicate
occurred. The
flip-flop is reset by the
condition.
computer program when it recognizes the overload
A
set
of level
that
discriminators
critical point in the synthesizer.
is supplied
for each
Currently fifteen
are
used.
2.10.2 Higher Pole Correction
Fant
(2,11) has shown that if the transfer function of
a uniform
straight
including
a finite number of poles,
tube
is approximated
by
a function
there will be an error
in the level of the high frequency spectrum. A uniform tube
closed
at
one end
and open at the
-26-
other has
a transfer
function containing peaks at all odd multiples of the lowest
approximation contains the first
four poles (500 cps,
2500 cps, and 3500 cps), the
cps,
4000 cps.
1500
at
error will be 41 db.
not compensated, a
this error is
If
500 cps and the
resonance is at
If the lowest
resonance.
noticeable
degradation in quality results.
The higher pole correction circuit used in this synthesizer was developed for an exploratory synthesizer (6).
higher
spectrum, and correct
shape the glottal
impedance.
If
the higher pole
these other circuits,
CCP
pole
and
radiation
correction is isolated from
zeroes
transfer function obtained matches
to + 4 db.
for the
the transfer function
real axis
two
other circuits to
is combined with
pole correction
The
consists of
(figure
a
The
2.18).
Fant
the curve given by
up to 4 kcps.
2.10.3 Output Amplifier
The
output amplifier
different
parts
impedance
output
equipment.
mixing
of
the
to
serves to
synthesizer
is accomplished
emitter follower drives
and
a
low
and monitoring
shown in figure 2.19.
by the input
the output
The
resistor network.
transformer whose
anced output appears at the output connectors.
-a?-T
signals from
provide
recorders
drive tape
The circuit used is
mix the
An
bal-
Chapter Three
The Synthesizer Control Program
3.1 Overall Program Description
The
program
is logically and physically divided into
six parts.
One
permanently
and contains routines common to all other parts
of these
parts
resides in
plus all of the
variables which must
part
The other
is run.
five
core
storage
be preserved as
parts reside
on
each
the drum
auxiliary storage unit and are
read into core storage
when
they
of
as a
are
procedure.
the
to
be
run.
Each
parts
runs
That is, there is no modification of any part of
program while it runs.
Thus, when it is finished there
is no necessity to save it.
be
these
The next time this part is to
used, a fresh copy is read
from the drum.
This saves a
drum swap every time another part is brought in.
The terms input data and output data as used here refer
to lists of numbers each element of which is a sample from a
time function. The entire list specifies the time
for
the duration of an
utterance.
function
The distinction between
input data and output data is that output data lists contain
numbers
which are directly
transferred to the synthesizer,
-28-
whereas input data
is entered by
the experimenter, and
is
never directly used to control the synthesizer.
The
five
parts are
three, buffer display,
named
and editor.
between
them is such that no
occurs.
The
follows.
tasks
part one,
The
part
division of
assigned to
these
five parts
input data
into
výa the
Part two performs
output data
editor.
generated by part two
This
are
as
Part one is responsible for accepting, generating,
the other parts.
entered
labor
appreciable program swapping
and manipulating input data and coordinating the
of
two, part
according
Part
activities
the translation of
to a
three uses
text
string
the output data
to actually control the
synthesizer.
part also monitors the overload detector and timing of
the whole operation.
The editor
is available to enter
and
alter
the text string
tion.
The buffer display is a short program that displays a
specifying the input-output transla-
graph of the current input data.
3.2 Description of Part One
3.2.1 Commands and Data Storage
All
action within part one is initiated by typing in a
command consisting of a single lower case letter or
-29-
special
character which may
arguments.
and
be preceded
The number of
some
commands
1U,.
things
a comma
starts the next.
command
depending
on the
An argument is composed of one or more
integers separated by one
Typing
number of numeric
arguments depends on the
do different
number of arguments.
decimal
by any
of the operators 0+0, or
terminates the
current argument
and
Argument handling commands allow errors to
be corrected anywhere in
the string as
well as wiping
out
the whole string.
The main data storage is provided in a number of tables
on the magnetic drum.
drum field (consisting of
Each
4096
18 bit computer words) used for data is divided into several
number
The
tables.
tables
of
typed in when the program is first started.
parameter
element of a table is associated
time
in
by
is set
field
per
the
utterance.
with a particular
Each
sample
specify the
elements
Successive
a
sampling
value of a function at successive multiples of the
of
a time
function., Nominally, each drum field represents one
utter-
Thus
interval.
ance,
i.e.
the
each
table
tables
on
contains
that
samples
field
contain
functions necessary to specify one utterance.
however,
In
all the
practice,
this arrangement is only necessary when that field
is to be used directly for an utterance.
-30 -
In the process of arranging
produce
an utterance,
data from
one
provided
to
to move data
also make
various
table
it often
it
manipulation, the
field is used.
becomes necessary
another.
around on the
possible
data tables.
all of the data needed
to examine
Several
drum.
(on
to
to move
commands
are
These commands
the
CRT
display)
In order to simplify much of the data
concept of
a current
table and
current
Manipulation commands which do not specify a
table or field use the current table or field.
Thus if
all
the data being maneuvered is on one field, that argument may
be omitted,
and
the
current field
be
This
used.
necessary and the possibility
the amount of typing
reduces
will
of error which it entails.
3.2.2 Data Input
Data may be entered in
on
three wayst
the CRT display with the
fitting to
points
entered
by drawing
curves
light pen, by programmed curve
from the
typewriter,
or
from
preparedpaper tapes.
When
the light pen
displays a pattern
light
of random
pen sees one of them.
to the four
From
input mode is active, the program
extremes of
the x-axis average
points on the
of these points,
-3'-
the
The program then draws a cross
the light pen's
coordinate which is a multiple
CRT until
field of
vision.
the closest time
of the sampling interval
is
found
and the
average y
coordinate replaces
entry for that time in the core data table.
the previous
Irregularities
in the input are smoothed by passing the x and y coordinates
through programmed single pole
x and y were
both
though
with
a time
time functions)
sampled
constant of sixteen points.
low pass filters (as
This filtering is applied
only
to the points seen by the pen and not to data already in the
table.
The curve fitting routine uses the LaGrange
tion
formula
polynomials
(21) to
to
input
exactly fit
points
segments of
from the
interpolan-th order
typewriter.
The
equations used are given below.
Let
=
Then
The
Xi and Yk are
iLT
Pk(-)
VP
('-A--Ai)
C) /P.(%-)
the n points straddling X plus one more
point fixed by the program
to produce a continuous
deriva-
tive at the transition between segments of the fitted curve.
The elements of
the sum are computed
fixed point for speed.
in single
precision
To avoid an intermediate overflow or
underflow, the numerator product is carried to the point
imminent overflow, then a division is done.
tions are continued in this
of
The multiplica-
way until there are none
left,
then the remaining divisions are performed. When one of the
x coordinate passes one
of the Xi,
-321-
the range of
specified
points
used in the
fitting is changed
so the interpolated
point is as nearly in the middle as possible.
For polynomials of
used
to draw
points.
degree one, a
straight line segments
separate routine
is
between the specified
The formula used is
-
L
ýj
1)(V2ý-19
(-A
Where XI, YI, and X2, Y2 surround X.
In all cases
the
curves are drawn only between and including the most extreme
•oints speified.
Paper tape input is done by reading binary data
from
which data is
table.
program.
loaded point by point into the core data
The primary purpose of
previously
blocks
dumped data or to
this input is for
reloading
enter the output from another
At any time,
a
the core data table may be smoothed
smoothing command.
the core data
range
table which are
through a non-linear
used
in delay
currently in the
horizontal
low pass filter.
+
r- _
O,_,_I_, 4
WAhere
O., S tke
Cf)
n
ouIPt
15
for
x
-16 for x
use of
The
equa-
notation is
On=
and
The
those points in
of the display (as determined by display manipulation
commands)
tion
This conmmand passes
using
barmple
> t6
the clipping function i'f" is
effect of any artifacts
in
-- 35 b -
t
O
)
point
)-t6
to minimize
the
the
data.
By including three points in determining On, the
effect of time shift inherent in simpler low pass filters is
minimized.
3.2.3 Data Display
while sitting in its listen loop waiting for
Normally,
type-in, the program displays the contents of the core
table
in graphical form
includes a fully
on the CRT
display.
labelled coordinate grid
data
This display
plus data.
The
grid is calculated to have major divisions of 5, 10, 20, 50,
100, etc.
the
and minor divisions of 1, 2, 5, 10, 20, etc.
As
size of the display is changed, the grid is adjusted to
always yield a
reasonable display.
The
current table
and
field numbers are also displayed for quick reference.
Commands
direction or
This
allows
detail.
are
available
to change
any
to move
the vertical
portion of
the
display
in any
or horizontal
scale.
the
data to
be
examined in
These commands affect only the display and not
data itself,
the
3.2.4 Calibration of the Synthesizer
To
facilitate any necessary
calibration or adjustment
of the synthesizer, two commands may be typed into part one.
One
command causes 18 bits to
word toggle register of the
data.
This
causes
a
computer to the synthesizer
is useful for
generating steady
tones
comparison
be transferred from the test
balancing
for signal
of the
timing
the
as
attenuators or
tracing.
of the
The
second
clock
in the
synthesizer to the computers timing.
This was intended
setting
but seems
to indicate
only that the computer is 5 percent slower than
advertised.
the 100kc
clock frequency,
for
3.2.5 Other Features
A
command
is available
first glottal pulse
used
to initiate
for setting
to within 10
other parts
usec.
of the
the time
Commands are
program such
of the
also
as the
editor and part two.
3.2.6 Error Comments
Whenever an inconsistent set of arguments is given to a
command,
an error comment
is typed out.
The comments are
abbreviated to five letters such as "ildeg" (illegal degree)
to
avoid the time necessary to type out longer comments and
to conserve storage space.
various types of errors.
errors and
About
if impossible
data
detected,
error).
situations
There are
arise,
The computer
from
the drum and
the program prints out
conditions
no halts
the
comment
performs a parity
if an
error
is
the comment "drerr" (drum
in the
program.
All
unusual
cause an error printout and the program recovers
and continues
debugged
read
for
The program also checks itself for
"'program error" is typed.
check on
15 comments are used
to
running.
be able
The program
to
is now
attribute most
sufficiently
stops
or program
errors to malfunction of the computer.
3.3 Operation of the Editor
The mapping specification
text
string.
the drum as
data to output
To allow
data.
text in the buffer may
and
punching paper
specifications would be
Lines
be examined, deleted, changed or
inserted using various commands.
reading
rapid
to this string, an editor patterned after the PDP-1
Expensive. Typewriter program (26) has been included.
of
a
This string is scanned by part two during the
translation of input
changes
is stored on
Provision is also made for
tape.
In
normal
prepared in advance,
would not worry about what it consists of.
use these
and the
user
Internally,
the text is represented by a character set
consisting of lower case letters, numbers, and the
tion + - ( ) = x V A ~ / .
The necessary
input
conversion
and output.
to facilitate
, tab space and carriage return.
is
performed
The different
the
punctua-
operation
automatically
for
character set is employed
of part
by
two
classification of character types easier.
making
the
The conversion is
done in the editor because space is more plentiful.
The commands
preceded
to the
editor
are also
by arguments as in part one.
single
letters
Unfortunately, there
are not enough letters to accommodate all the commands,
the
so
same letters are used for commands to both part one and
the editor.
editor
To help avoid confusion, commands typed to
are printed in red and
the
require a carriage return to
be typed after the letter.
3.4 Operation of Part Two
Upon command to part one to produce an utterance,
two
is called to perform the input-output translation.
details of
text
this translation
string which
through the
data
part
editor.
tables, other
are determined
was previously entered
In
addition to
ance.
A_7-
by scanning
a
into the program
producing the
information is prepared
The
output
for the utter-
2
r"
The output data is first compiled onto a drum field
the
same format as
the input data
used is not available for
tion
on
output
this
tables.
The drum field
alloca-
input data. The storage
field
is determined
in
by
an output
variable declaration line in the beginning of the specification string.
consists of a series of
body of the specification
The
algebraic statements written
in terms
of input
variables,
output variables, temporary variables, constants, and operaEach statement
tors.
is applied
to every
the
element of
variable tables.
Variable names consist of two parts, a letter (i, o, or
t for
input, output,
variable,
followed by a number identifying
Thus 12 refers
output
denoting the
or temporary)
to input
2, o4
data table
data table whose contents
type
of
the variable.
refers to
that
will eventually appear on
channel 4. The storage for-temporary variables is allocated
automatically
by
the
appearance
of
the
variable
name.
Storage for temporary variables and for intermediate results
is provided on
purpose.
scaling
drum fields
not being used for
any other
All arithmetic is done in floating point to
problems.
avoid
Input data and output data are converted
to floating point for the calculations, and then returned to
fixed point.
3.4.1 Expression Evaluation
into three
a full
is done using
The evaluation
parts.
is divided
the expression
of evaluating
process
The
priority push-down scheme which manipulates
operators and
pointers to the data tables feeding them to a second routine
does the actual arithmetic.
which
names
variables
function
string assembling constants,
scanning of the text
names,
A third routine does the
etc.
providing
and
pointers and
operators to the main routine.
The evaluation is defined recursively as follows.
The
routine evaluates the expression up to but not including the
operator whose precedence is not greater than the precedence
of the operator which called it. When it can go no further,
the routine returns
of
results
the
a pointer to
the table containing
sub-expression just
The flow
evaluated.
Note
diagram for the process is illustrated in figure 3.2.
that
the
scanning process
does
not penetrate
a carriage
return and effectively causes repeated returns to close
extra
open parentheses and
finally return to
called from. Unary operators,
ceded by another operator,
of priority.
Omitted
multiplication.
the
all
where it was
distinguished by being
pre-
are done immediately regardless
operators are taken
to mean
implied
That is, 12(ti-3) is taken to be 12x(ti-3).
Operators on the
except for "="
same level
left
to right since this
age.
Equal
sign
are done
from
generates less intermediate stor-
is done
from
right to
left
to
follow
conventional practice.
3.4.2 Special Events Compiler
After
completing the
the text is scanned to
output compilation,
compile the
the rest of
special events
table.
The special events table is used for changing the configuration switch settings during the utterance.
"at
0,100, set 1,2
reset 3'
are
Strings such ast
processed.
For each time
specified, two words are stored in the special events table.
The
first
gives
the
time,
the
second
contains
a mask
indicating in one half, which switches are to be set and
in
the other half, which are to be reset at that time.
The
algorithm
for decoding
these statements
is very
siple-minded.
The flow chart in figure 3.3 should be self-
explanatory.
Note that
the first
used to
distinguish them.
anisme
would be the same as
well as any other
Thus
"at".
letter of
the words is
"antidisestablishmentariAlso space and comma as
non-numeric and non-alphabetic
character
serve only as a separators, and 20 separators are as good as
one.
The intention is to
provide an asily readable
with a fairly obvious meaning.
format
When
all the text has
into half
on the drum are squeezed
That
is, 409 6 words
As
words.
this
been scanned, the output tables
are packed
is done,
their original
two to
the range
a word
of
the
space.
into 2048
numbers
is
Out of range numbers (which cannot be transferred
checked.
synthesizer) cause
cating
in which output channel
is necessary so all
print-out
an informative
to the
they occurred.
the data can fit
indi-
The packing
into core along with
part three.
3.4.3 Error Comments
translation, the
performing the
While
program checks
for certain syntactic errors such as illegal punctuation
illegal function names.
such errors
positive
as division
number.
The arithemetic is also checked for
the
or logarithm
by zero,
If any such error
error was
found.
of a non-
is found, a comment is
of text was being
typed out indicating what line
when
or
No
attempt
has been
distinguish between various types of errors.
-4'-
processed
made to
3.5 Operation of Part Three
The third part of the program takes care of the
transfer of data to the synthesizer.
actual
The cycle of events in
the main loop is as follows.
I) Set up the data.
That is, unpack it from the table, tack
a channel number on its front end, and put it into the
output list.
This is done for each output item that is
to occur at the next sample point.
2)
Wait
for the
synthesizer.
right time
to transfer
(The sequence break
the data
to the
routine is keeping
track of the time.)
3) Transfer the data and return to step one unless there is
no more data,
4) If there is no more data, reset the data pointers,
delay
for a while and return to step 1.
The
sequence
break
hardware clock times out.
routine
is initiated
It does the following.
-42-
when
the
1) If a glottal pulse is to be generated, do it.
2) If a
glottal pulse will
occur within the
range of
one
countdown of the clock, go to step 8.
3) Set the clock to delay one millisecond.
4) Add the amount of the clock setting to the current time.
If it is time for more data and the main loop is ready to
put it out, signal the main loop to go ahead.
6)
If the
main loop is not ready, type
out error comment
"too fast" and stop the utterance.
7) Return to the main loop.
8) Calculate the interval to the next glottal pulse, and set
the clock for that delay.
glottal pulse
break.
9) Go to step 4.
is to
Set a flag indicating that a
be generated
at the
next
clock
The
glottal pulse
quency data at
next
pulse
every sample
is adjusted
interval between
After
timing is determined
subsequent
the
The time
new
glottal
left to
frequency,
pulses
the
and the
is computed.
each glottal pulse, this is used to determine when to
produce the next pulse.
until
to
point.
from the fre-
after the first
this t-o
frequency
be
precisely
glottal pulse is generated to allow
specified.
In
the
case
of
zero
pulses, a flag is set to indicate that regardless
of any numbers
(except
The updating process is inhibited
the
which may
be calculated,
first) is to be
generated.
no glottal
pulse
As soon as the
frequency becomes non-zero, the flag is cleared.
While the program is waiting for the go ahead signal to
transfer
the next data, it checks
have been typed on the typewriter.
are
taken as commands to either go
to see if any characters
The characters so
back to part one, or to
stop talking but not go back to part one.
is also a command to start talking again.
- 44--
typed
Naturally,
there
Chapter Four
4.1
The Scope of the Evaluation
The
evaluation
It
parts.
as
of
the
synthesizer
is important to know how it
well as how good the speech
performance can
ratio,
it
Performance Evaluation
be measured
comparing
of the
Technical
of signal
to
noise
The quality of the speech
judged through inforrrmal
spectrograms
of two
technically
performs
produces is.
in terms
distortion, and accuracy.
produces has been
by
it
consists
istenng
synthesized
and
versus the
original speech.
In addition to the
another
measure is how efficiently
the synthesizer.
nity
above two measures of
performance,
an experimenter can use
At this time, we have not had an
opportu-
to let many people try their hand at using the device,
so this aspect of
its performance can
only be judged
the author's rather prejudiced experience.
from
4.2 Technical Performance
4.2.1 Pole-Zero Pair Circuit
The tests were pole frequency
was performed. on it.
dc
accuracy,
imaginary parts),
real and
accuracy (both
over several
offset
to more
extensive
of the resonant circuits,
problems than the rest
testing
prone
is
pair circuit
pole-zero
the
Since
zero
frequency
settings and
frequency
ambient temperatures, and interchangability measurements.
measurements
The worst results were obtained in the dc
at
At
temperatures.
elevated
much as 3 volts
settings,
low frequency
of dc offset appeared
at the output.
as
The
cause of this is principally the low loop gain combined with
offsets introduced by the
The feedback in
the
tends to compensate for these offsets by offsetting
circuit
the output.
The low
gain in the
offset to effect
large
attenuator.
Fortunately, the
applied
to any
feedback path requires
a small compensation
output of
the
at the input.
pole-zero circuit
rapidly switching circuit.
a
is
not
The results of
the dc tests are tabulated in table 4.1.
The frequency accuracy tests were made by measuring the
frequencies
at which the amplitude response was 3 db.
from the resonant peak (3db.
zeroes).
The
average of
up
these
- 46-
points in the case of
was taken
as
down
the
the center
and
frequency,
the
divided
difference
two
by
the
as
Both the center frequency attenuators and band-
bandwidth.
and
width attenuators were varied to determine the accuracy
the
orthogonality of
of these
The results
the controls.
4.3.
measurements appear in graphs in figures 4.1, 4.2, and
A comment about the center frequency accuracy curves is
the quantity which is
In the circuit,
necessary.
actually
set by the attenuators is the so-called natural frequency of
the circuit
,,,,=
,X]
-cr?
*-++
wn.
The curve which depicted is not
where w and wn are
damping
greatly different, a high error
The curves for smaller bandwidths more
results.
high
As a result, for
accurately
reflect the actual performance.
bandwidth accuracy curves in figure 4.3 indicate a
The
rather high error in the bandwidth of the zero.
to
an error in choosing
the zero bandwidth trimming resis-
The measurement of the
tor.
chosen on the
basis of incorrect
was re-trinnimed,
measurements.
3
db.
points
is
After
the
the badwidth accuracy was improved
to be about the same as that of the pole.
the
very
bandwidth of the zero is
due to the low signal level, and the resistor was
difficult
resistor
This is due
not
an
accurate
Again the use
measure
of
of the
bandwidth for low Q, and the error at low frequencies is not
-*47-
accurately given by the curves.
The
interchangability
tests
revealed
that replacing
operational amplifiers or attenuators caused essentially
change
in the ac accuracy.
The differences that were noted
are completely accounted for by experimental errors such
meter
interpolation.
The
no
dc performance was
as
upset due to
the external balancing adjustments on the operational amplifiers.
After readjustment, the
performance was equivalent
to the original.
4.2.2 Formnt Ci•rc•uit
Since the formant
this
thesis, no
extensive testing
Some measurements were
and
are reported in
expect worse
circuit.
direct
circuits were not
has been
made when the
reference (6).
performance than
built as part
done on them.
circuits were
There
that of
of
built,
is no reason to
the pole-zero
pair
The dc characteristics are much better due to the
feedback
from
output
to
input.
The
offset
is
typically less than 5 my.
4.2.3 Interconnecting Switches
The
results of measurements on one of the interconnec-
ting switches are tabulated in table 4.2.
4.2.4 The Overall Synthesizer
The electrical tests nmade
consisted
chiefly of
on the complete
noise measurements.
synthesizer
There are three
Random
principle sources of noise in the attenuator.
is
generated
by
the circuits
themselves
and
rushing sound with a tone-like quality due to the
action
of
the
resonators.
Non-erandom
by
is
4y
induction (usually magnetic) from
one of these sources.
the iogic
si.nam
of the
s from
noise is
Hum caused
the 60 cps power line
t~he computer.
digital attenuators
caused
These appear as
The third source is
noise exist in this category.
crosstalk
filtering
The other is ground shift
varying high pitched sounds.
switching
produces a
induced
produced mainly by ground potential differences.
noise
(7).
caused
by
Three types of
Additive noise appears due to
of the digital signals (due to capacitance in the
switching transistors) into the analog signal paths.
Multi-
plicative noise results from modulation effects of switching
the analog signals.
The third
type is caused by
switching
the unwanted dc component in the analog signals.
The
measurements
indicate
the
random noise and multiplicative noise are insignificant.
In
the
static condition
major offender is hum.
utterance
and
tests
(no connection to
With
being generated,
from the changing
listening
the computer),
the computer connected and
the most audible
logic levels from
- 4-9 -
the
no
noise is that
the computer.
If
the
h
s
i
a
d
ha
a
h
b
t
t
excitation applied, the switched
The results of the
nant.
4.3.
in table
The
dc noise becomes
noise measurements are tabulated
is the
most interest
figure of
noise
Compared with a typical
during speaking with no excitation.
vowel /o/,
predomi-
the S/N ratio is 35 db.
4.3 Listening Evaluation
The first utterances which were generated, were
The data for generating these sounds
vowels and d.iphthongs.
were taken from references
as well as
(13,14,17),
of the author saying the
grams
simple
sounds.
spectro-
A wiring error and
inaccuracy in the spectrograms contributed to a poor result.
After
fixing
the wiring
error,
specification to compensate for
a very good result
trograph,
and /j/
were
Spectrograms
also
of the
produced
and changing
the mapping
the inaccuracy in th
was obtained.
with
equally
synthesized speech
spec-
The glides /w/
good
results.
are illustrated in
figure 4.4
The next
connected
speech.
corded and a
was
step
in
The
difficulty was
the
production
sentenced "where are you"
spectrogram made.
entered into the program
Data
from the
of
was re-
spectrogram
using piecewise linear curve-
fitting.
The result was good.
synthe-
Spectrograms of the
sized version are shown in figure 4.5.
The
stops /b/, /d/, and /g/
were produced next.
These
consonants can be characterized by a rapid transition of the
formants (typically in 35 msec.)
The thirty sounds
(15,18).
produced by combining the 10 vowels and diphthongs /i/, /e/,
/a/, /a/, /I/,
stops in the
test
/o/, /u/, / i/, /ay/, and /aw/ with the above
initial position were
data for
synthesized to
produce
The original
data was
another experiment.
taken from spectrograms of the sounds and augmented with the
results
of other researchers
(13,15,17,18), and then hand-
tailored to produce the final utterance.
obtained
with
the exception
successfully synthesized.
of
Good results
/di/ which
A representative
could
the sounds, that
not be
collection
spectrograms of the results are shown in figure 4.6.
noticed in producing
were
very little
of
It was
"thump"
(which has been noticed in other synthesizers) was present.
The
most
recent
utterance
produced
containing voiced stos, glides, diphthongs,
sentence is
this
"are
you a good
utterance were
the curves
for
boy, or a
taken directly from
the four
formants
is
a
sentence
and vowels.
bad boy?" Data
The
for
a spectrogram and
and glottal
pitch
and
amplitude drawn directly with the light pen input on the CRT
display.
It required 35 minutes
- 61 -
to enter all the data
and
produce
in copying part of the
good but there was an error
frequency
After a
sentence
The results were
the first draft of the utterance.
curve and the
few
was
more trials,
a
produced.
The
utterance was 45 minutes.
was not very clear.
in "good"
/g/
very good
total
glottal
synthesis
time
to
of
the
prepare the
1
F1
-I
1
I
I'
hi
---
I
I
I
I
-
---3Ufe~L~.~~6EYS-1-L
~
CI~
I
I
1
J
~·iJ-crC
i
Figure 2.1
Block Diagram of Complete Synthesizer.
3'O
Jeis;~.K
Figure 2.2a
Analog Computer Simulation of Resonator.
Racnrr
~
0
477-
avJ = 2? c
Figure 2.2b
rtao~c
1
Rc
=- CRI
--L
Realization of the Resonator TTsing Operational
Amplifiers .
e04
Figure 2.,4
First Realization of Pole-Zero Pair.
Figure 2.5
Modification of Figure 2.4 to Allow
Pole Frequency. to be Less Than the
Zero Freauency.
Figure 2.6
Block Diagram of Final Pole-Zero Circuit.
in
Detailed Circuit for Pole-Zero Pair.
Figure 2.7
Notes oil figure 2.7:
* R3
is
of
is
of
augmented with a selected resistor to trim bandwidth
the zero.
augmented with a selected resistor to trim bandwidth
pole. (R4 must be selected before R3)
R2
is
of
is
of
selected to trim the maximum resonant frequency
the- zero.
selected to trim the maximum resonant frequency
the pole.
Cl
C3
is matched to C2 (not necessary, but a good idea.).
is matched to C4 (also not necessary.)
* R4
Rl
All resistor are high quality 19 1/4 or 1/2 watt carbonfilm.
C1-C4 are high quality silver-mica capacitors.
a0 , al , b0
,
b1 are digital attenuators.
Figure 2.8
Simple Resonator Diagram.
jW
Doc0t
[ec41.1
--
I
is
.. . ..
.
. .
.
..
.
open loop
Figure 2.9
Root Locus Diagra. for
Simple Resonator.
Cý
Figure 2.10
__
Differential Integrator.
.
l
fnW
%_j
v
r
I
•.
f-
pla aCnd tvro
Figure 2.11
Pole-Zero Plot for Non-Ideal
Differential Integrator.
P
10
Figure 2.12
w
--- *
Phase Angle of Non-Tdeal Integrator.
I
1
1
_--
-
"S
r 1
c1
entyo;d o-open loop
Spoles
Figure 2.13
and zeroes
Root-Locus Diagram for Resonator with
Non-Ideal Integrator (no damping).
.1"
o
Fig. 2.14- Determination of Phase Shift due to
Extraneous Pole and Zero
i
(a)
A=
Ba.sic A4+en .o.+or
+6
R:
Pvr inciple
-15
(b)
4-
DC SALANCE
OUTprU
INPUT
+15
Ty p i co
Fig. 2.15
A 4 + e n ui.
Co Tra O
2o-r
S a. e
Attenuator Stage Operation
w
- 16 V
+15\V
+
V
30 7
L fN
zoK
75K
~YOT1~
INPUT
0t
2No
13ý,
-.6V
2OK
7. K
- SV
+15V
Fig.
2.16
Interconnecting Switch Circuit
OfCF~
I
.
-
,
3.3 K
ZN 1306
C L'
(1rv)
(- v)
4701
Z.4 K
2N 130"
3.3 K
Fig, 2.17a
.02 mC-
Level Discriminator (Overload Detector)
-1•V
K.
Fig. 2.17b
Flip Flop and Indicator (Overload Detector)
-4
0 1I
3K
nputs
k-io
oI
-Ill
1
+15 V
Figure 2.18
Output Amplifier.
,v
balao.ced ou-+pu
(soo nC)
ENTRY
l AIN PROG.
"To vAL.
is the operator
whose priority is X.
is the priority o.f
operator X.
means carriage
return.
is the N-th syllable in the
expression.
means put X in the
next free entry on
the pushdown list,
Figure 3.2
Pushdown Evaluation
Flow Diagr.am.
POP'X
restores X from the
last entry on the
pushdown list.
CHR(N) is the N-th
character in
in the
string.
Figure 3.3
Special Events Compiler Flow Diagram.
a
ý
- ....
. . '
.-.- .... . ^I __Y_ __
ý
..I
. _ý4_ýf
._ -.
.
.
-.
.
.
..'.-i- .Z.'-_'-.tt';-ý--:_.-;._.......
...
ý
.
-'---j---j
t
'
'
1
..
'.
''
'
.
I
.....
'
.
L
i-4
ýtlý7t;_. .,:=_;A'-,1-:ý
-...
,=
,.:ý---.ý.t,!.I,ý--T.-;,ý-ý
_...
....
7,'
V.. ý__-_
7-- ___ !-. ..-.ý. :;., I... I~ ..- -'--;,
a I .- .
- _1
_
.j__
'I-: .ý
...
.-_ý'.
.,
.4_i_
:ý.'.".--;-:YýlLý---=,I-,.-;t-'
....
t
ý
t
ý
:
ý
±
.-.
i
_1
' !-;
ii-i
ý
.I.--_
ý
.
.:.
..
1;
--F
.
.
.
..
I
'...
ti
_!____j._
-_
.
,
-1
r'
"_
__:
_a.ý
_ý
....
ý
.4-1
--1
-1
.+
...
..:
....
L..
I
I
_t.
-_
-,
-1
-_
.
-.
t--.!.1
-.;
.
t
;
ý
--1
%_
1.
.
.
.
M-4.
--ý
::I
- :-4a I 3 _- .ý_;.- - ý__;4-I
. . . . -!
. I ., - - ý -- -_-.-__' -ý-___'_ - -..-.-.-;*4.
..
ý. I.:
I ; .. ...
!--;
-ý ! -.1.
4'__-fý
m-- _'- -:
, _- 4 __; - . . - ... - . -I ý."-I-'ý .. - . _..... -ý. .1.....ý- ,.-, :...'....I..
'. I;
-17
i7 ,--.-.;.
-1 ...
: L.i - ; ___06-.- -L
-----
7-7...,
.
.-:=ý7-_. "-.;'j.!---ý.ý-;.-t."_2-7---
__
ý
..
.. a
. ..II ...;-ý .-:...
:..,. 4
-i
...
"I':
_;.
..
I
. . I_m
...
....
j_.I_-_d._'-.-...
I.-..-ý--ý i-..".. "__4
7._-,:n_
Ji_ý_i
...
ý;_;-ý_!_' , :ýi.' :ýLý::
:-,-.:ý, .
; .. ý::
---- i I . .
. ; __
, . _"---i-41-ý-;-:-1--ý-"-t-1ý-:
ý-""
-------ý ,,-;
. !...-.
: ý'.
__ ..
.
; - .
. . . ý
. I...
;_ .... ..I!_.-.. ;-- -111-- -ý...;.
- -.iý..
,: .' .4_ý
_ .2::,.ý...
ý - ý, 4I- .
., ...!- . '.ý ý..:ý'_.__'J-__.
L
ý4-..I.-.--.::.'-:,-----,..;--:-ý---.1-1...
_ý'-ý_-ý....
:--__,___....
i,- . T..... I .. ! .'. . 1., .I.17. r...-...;-,..- w.- -, .
.-:-ý-ý-ý ý ._ý __-1 I_.
I..,-.;
.1..._ý..-j_
:
ý
Lj::;ý:1._l_-...
_.E.-ý___
ial
t
:.
ýý_ý.-.--.-.-'
-_ýt
-Lj-t:J'-:t
z
..I.-I
;
..
:
_j
-_-.I..
,;-.!
.1
-ý'
ý
ý_. ....i
. _. ý-;i.
' .ý.. .:.-:..
. .a..ý. _m
..." .-I. _.
. _1
.
L
ý
.
.
t
.
,
..Iý ......
:. . .........
. ....
I
...I.I.
_...
-.......
_ý_
ý
-i
-.
-,
j
ý4
._
-_:.
...
..
-:
:
.
'_ý,
-:..
ý._-ý-'
4.
ý.__;
4-4i.
.
"
."...4-ý;
r.1
-.-:
-ý-ýý-4
.
'...'
I
.
.
-1
-I.
.1.- -.- -'-'--.ý..+-_L
I ý .I . ....
. .'. ;-I.
,- -"--L-tl
ý:'.
4.
..
.
;
"
-'!
.I.
...
:..; ;
....
. . . -ý
I i.' .: -..
__.
- .- .-, 4- 1_1
, -ý
47-1
J_.'..
--.----.
-ý-7-' -i-; --Lý_ý
- ....-- . - .I -_ý-.+4-_.:_L i.-_;_..-.
....
I
;__L
I
m...
,
. _.:I . ' ......I....i '-1-1 T_
..:.I .:_.
I : .
I . . .__ý;-4+ ;_,--l"ý_'_._- i-i-...F'--__Lý_;..;._ý_l'..'__:.
.
+-7- -ýi- i-4-.'--ýý-;
-ý- 4- 4-4.ý__r - - . ý_ ý -ý.- . -1
, .-.._;__i
1
-:
70
ý_
'.
J
.--."
,
...
'-...w
...
'.
I
`
-.
.
.
.
: ; . .1 -. -. I ; .
'
:
. ..
. .
I . . Ia . . . . , -7,,-'-,_'
i
...
.
:
t
;
.
,
.I
:
i
_
;
.
71,__
. I
._'_ý
- _'_ý-1-_: ;J4.- : __."'
,
__'.ý_ L- Iý .;-- : i .
.....' _" '
_:..I._ýý_ý'...; i .:... ... I. .
. j 1.;
-i - 1..;i..... .I ! . T..
.L!_ý......'.__.i.._
- I --,:--, .-.-;-,-IJ___ý _!..4...
'ý..' . . ...
i.-ý.-L-I.-ý-'.-'--ý--,..---ý4-ý-.,,-L
-4..Lj.. ,
!. _ý
.-+- - - ..
. ._1
'- . I'--.-I ;_.ý._'
".ý_ -.7......-.-.
j...'_;..
I.- ý.- ..' .;...- 1..1. - . -t...
f !. 4.1
. - ý_
'. I
... - 1.
i..i--..+ý
- ..
L_.. t
...- _+ Lý__+4
_" -,-:.! ., -,
.
___
3
ý 1±
.1_;._..'ý:
Iý...
..
L;.
.-_1
14'. ý
_il_--.; 1 . _ý
-,_'__ý12
- ,--:'.'_:':'
-1+_
'7ý_ . !.!._._':'-..'_
-4 :. -";_'_'.=
l_.._'._.'p.
..
I;...... . ý...
...
"ý
-..i_ýi
:. ! :.
_.z.
':.
ý
-i
ý
-;
_I
,
"
'ý
.
.
-4.4
..
1.
.
.
._'
.L.A.
7_.;
Z.
.
:
ý
.
.
I
.....
.
.
.
,
.
..
.-.-.
..r
.
I
I
Id..
::
-.
:
_'
_!
-.
1
--i
:
z.
.
.
....
:
:,
:
.
..
;
........
I- _.ý-..
..
_
_
.1
-i:.:
:
.
.
!
_:..-_'_
;
----ý'-ý.'..t
.:ý
ý..
m
..
"
ý_'
T.
ý
...
....
ý
.
__...._...._-._
'I..
. ..... .
.l....,....
._!_.
.. _...
- 'I
t-ý- ý- _ - ----ý---; ...
. -1.ý_.;. ......
.ý.
..I..
. _ý. :... __ -.. .. ...: .- - __..
ý.
_..ý_
- _"..1 j__'..'...
;
. -.ý......
: - -.7_.i-_. 1.
. _...
.1-h:ý---. -_ -.1-!--".
. ...
.
'-- -, 4, .- ,. . .ý
. --!_ý
_.- 7I.:.;-."'Ll '-ý..:4 -:1-.7, .ý._
I -"
- - ý-::ý_"':::!
- I ý...t.I .
..
_.
I
1
. I
:. .
.
- . ..
'ý
` ,,
iý 444
. -.
.I...
___. '.-ý. A.= -. ..ý-7--..ý4.- . . .....
-j.. .:. ;. ... .. 4-::: :.1.I. . ;.-1.:.,..-,.
...
_
+
.
.
.
.
.
U,
ý
ý
I
1,
.
.
.
.
.
--,
ý
"
3_
",
:.':.
..;... - -. _;ý _-t- .1 .......' ýý..= 7:'7 -- .-.
.1 -.,I..
- _.4--,- _ý.
.- ._=.-.-ý7ý'Zt;_.--.:j::ýý
. __- . ý .
__ F__
. . - .. ..
. -;-. +- - L.
-:-i-Tf-+
-!- . -1.
ý]'-r; .1.
ýý:..:
........
--'
.!-.:
I_
...I-..
' ý__
I :-ý"-..f- ...ý__._
--- --_ : __ . - . I.
__
ý--.- ý'.
+..:
ýý'...
ýt1ý
.....
;.....
ý..ý._
t.i 1-1_;..!,1.f__t.--1-ý
:'ýý
,_-;--;,
I-1 . . . :i - -:- ..:.......
...
. .. ý
z-;_.ý
. _.
.........
;
...
_ ý '_i . .
. . ýI-;-ý:
.
.
.ý---; '"ý. tit., '=. Iý .- .. ! .: I ..
.I-.
---I .-. ý_;_-i
.- ý- __-. l.-.__l__.T
....4.. - -. _.....
_-i
......
_.- _.___
. I-Ii_;
- .,--...ý-.i.,'__
_.T_
- -I_1
_...
4--zl 4-4ý
.m." - . 1j.._... '_ý-----...--.....
_;
7-q-"p-- _fý_ý_.
. _ý'.
'_'- -_-:_--_-=ý::-.-v '_ý_.ý_!-_/.' - "- -.---. -_....
I . . w_:ý m
.I... ...
. -;l'Z__
z_-4--'ý".in
"'
.1
!-,---:;:ý-4ý4-,-,-ý,-,ýý
--...-.-.
__
-1
....
,.
;-ýf-!.
.'
-..-ý't r ,Ll
......
...
....
__;_
..
""
. - ..:m.
.ý'
_......`ý
'.. .t-_ .....
- _jý'..ý_
_t-.'-.l..._..l-_./
+LýL_ý_-ý. . . .. ,- .ý - . .. . . . . .,, .ý'_
_-ý%_..
..---.
ý-..-,ý.......ýý..i...,-;.---'_.:_.'_
-- l-_ .-- ;_.04 - ý...
I
:
Q
ý'.j
.
.
..
.7
;
i,
1-i
--I
..
r
..--_---4-,
.ý.a
.ý
.;
._L_
-;_.i__L_!
.-..
.
._.
:
:
-;-:
.
..
.
.
.z..- -; ...
::
:
- 4-.-ý -.: : " , -,.-!-" ; -, . .ý i-L 'ý_. .:.... .
.
.
.
.
I
.
..
I
,
i .j
4ý
.-.1 .
TIj ...
. .. . ;,
.ý-_
...-:.-. -ý
- _.....
! . -i- L-+.-,
-T
.
... ......
i ý.,.-..
.: . .-.... -I. ý .....
.-I
1: . wl
ý.---.__-"4
...j. :t ..-' ._. ...
..
t.....'
4K
4 ý'-:-ýý
j_.' ...
L;
. . _'....4
...
ý_._ý
-ý i -_
..- , - 2-ý. - ; [ . . . .1-; .2..ý__
4-1 ý__.:
.
.
! I ýý..ý_r'
. I. - -. ! _: II. ..I. . II ...-I .!- j...
; ...
-ýI..4..
-.:::...-.
-1
...
....
;;._i_.'.
'...__I -____ý_:._4__
-4- !-- - ..,.-, .ý
_ý__- -. . . ý
.:.-..-.
....
_;__I, ý_
.-.'_-.'
r 1 , i , -;- .4-j-,_-1
I- -1-_....I .- .: , . ..I ý .ý-----;. _i-_
.Z.ý
", .4--ý_; .A_..__'_.'_'..l
...
I7'...
,. - : ;_ý....
4:LA
..
7.
.
.1.ý.ý...... ._...ý_.;'
....
-...I-4--i
- 4 F-...-;-I,..-----:_'. 1.___ý
ý
...!_
i.-,ý ..:.,-"---.
- .. ..._t...x
ý_'.'
--I .P.._
I
__.
.' ýý . ý .' -T_!
I
_ý _;.---ý
.
--,
'--:
,
.4
:_:
.J
ý..;
,A
.
___
.'
.
ý.
ý.
r
'l
-.1
"-ý
..
!,
.
;
;
ý
ý-I.f_
.
71
ý
I
-.4-1-1-:
,
ý
___._
i
I
----_...'
;
f
_t
ý__
i
'..'
I
a
"
-l
_.7_:_i-_PlT
:
-1
.
, -' , .: ý 3 . .ý" ' . _,.-,----;...4
-ý:.
I.Ii 4--:
. . . .. I . i
q
. I
11
ý". -I. .:_L!_:
. .%-.:.
['ý.- .. -. 1; .;
.II. .i _.
.
I .
. : I.-. ,..__
! . .11.
,. !-- . - -ýý.I .. , m:.. . .ý__
._l_...:.__....;
.I f.
i I..
: I--.ai--;
- -1-'
.4..- ý 1--._ý
- ....
.__ .
. -1- - r. ý
....
so:.....ý, ._ý_jjý ,- .-1..- ...i.: ,-.,
...
: _-_: -..-;. ý__
:.-,..i_+-,
.
- --,!-- - :' -:...
-,; '-ý.".-'---ri
-. -. .- I--I ý_j.._Li_ý__-_i-F-.1-;ý- .-?...-,;."ý---.
..
...
m---.
;
..I..4-I.I
..
.
.
.....
.
ý
--..
.
ý-.
;
..
;
_.'.
-...
_....... .- "-.:
-.--....
.It-.-I-1-ý14ý-1e.ý....
...
.
i
. I-..i1.
4--l-4
....
-t-7--r-4.
_..
---.-..-..
1'.
,
,
I
:
.
,
I
.
I
,
ý
-----ý---. -,--- I
4.. 4 j- __ 1 ' ' ...ý , ; : .. . . . .I-I..,
- - ......
ý
- . -.
......
- -.I..I. 1--' j-. -44- --1....:ýid-.11.4
ý
!.-M.;
.
_...'
_j_
.4...l
I-_l
!.-!.
i
;_-- .--. - - ---__;..j_._-ýý Lý_,_:__
: --ý
ý_' .; ; .,
..
I
1.
.
.
.. . -o-... ý- -+--- -.
.
1-- ; .
. .
It
.
- r :erhý
...
..
-H! 7ý ,, . . .
.
!ý
J...
.... ..?'..- ..
, 1
i.I_:__ .ý...._.ý4 4_ý____L' __.Lý
:.- _...,_..ý
_4_4-, -- I . _:..
. 4... I ,-:ý.--t-iý
. . .- . : _ý'_:_'_[__!_i.__.;_'..
.
-_
_7- .....
..'.'-t . mI.- - - . - .....
"'
!
ý.___4.j..j._
r,-'
.
.
"--;-i
..i.,
'
_i-.!._._;..
d
i-Z-."_l_'l-;ý'_
.-L
F.__ ,.-."- . j_... ....ý......
....
'
_".-Ll_....-_ý.-'._ý_
_
;
......
-j_-ý_
;
ýý_;
'
.i ..
I , 1-1,
.
.:..:.
_i
:.
;
_ý..Lý.
.....
_ý-Ij-_2
...
1'..]ý` _.;. ý_. fA...' T_'.ý.'.f7i.. .4-4 - ý... m - ...: . ..ý..:
4, . I . : ! Lý...... ýAj ; .'.
4 I a__'ý..;_L _'._ý..
ý_"_...i'.' -. --- z.,
4.
.
. i.'j-.;."'.'. : : : t .ý .
_7'.I.~ ý
j ,-Lj-. L4I --:--ý-ý
I i I. : . ... - - . .
I I 1. . . . . .
.; .:.
.
ý
.-.- . '..'.
,i .- .I..4..f
, I._;__-it;1-. "
. : f ! I -_
! ;- +_;.......
t'. - T-r-[
1: , 1.
: _._i___.'_ý..
. . .I ,.; ;.
_Q_'q
'.
ý . ;
! . .i
. . .'_Ai__.
: . . I_.ý_!J.
_.- ...
ý....
, ..- t;ý
'.
-.:-_ý_!.-;-ý
4
:
ý
i
-i
_ý
.
.
1
-I-+_.":_..7-'..
'.;......
..--!
:.:-..i
1_ý.I : ,.. :.ý
.
.
.
:
:
'
,
,
.
.
I
.
ý
-ý
;
ý
7
; . ai
s
.
:
.
..
; !, , : :i ,
'_4"-.j- L- .- __!.:
- .:...'.
i
- .. , ; IL_
! . I; , . - A..:.-. : !A
f
.
:
.
I
I
t-r-;-.'
':_,ELL!__ý+._'_'.-;
...
;.-'-4._
.1
,
...
.
_.;
-6.
__
.
.
...
.
...
:
,
_V-ýJý-Ll_!.._._
.
- I j' . .. . Fý .. i . ; a I. .I. ý .-I
.. __
.:....
. +.: _+_'_
T-ýJ'_ý
. _-_!. .1_i_! _14._
7. 7ý i i I: t. .... . ý. . ! ; I r;;. .
-"
Ii. ; i. I I, .; . .
.. I- I ' A..
I
" __ ,. ...! .-,.,,
.
:
.
i
;
-l-L
A
1._,_!'__tý_Lý
'
,
ý_:
-.......
Lý_.
I
.
mI -IT-7
.
-ý-!-J-4-- -ii .
-; ,ý- ' : !
.
- ....--l'......'r..
.;_
,
._.., . I , ; . f- - ' . .-77.
, I
, i; .
i ..'.-..
!
_ý_._L
-,.'- .... .-.- ý_ý_ -_; . .
: ..ýý-'
.. ..i....... .:-V...
I-- ý . i- .L::::: ý, I Ii , -...
H_;1_4
;
:-. .
- _ý'___
'
'. .i-!- :-Z1.7-t-".; :i_;'
, . ý. : ;_ý..-_...l_.L.'. tT
:, -. _:
:
. . ---. .....- - 1 , __ýý
..-ý-.....
-,......
Lj_4..a .
:U. ý_ý____L.
. %T ..I
''
'- 1 1L_ýi_-;' -ý_-.A.
....
I I- .... . ___
'I
.
-II I----I-,I-I I ý-ý
. . _ ;-.:.."..; - ---r
.U.
r
ý
-. I.,
...i-.-'r ---. ...ý_.L
....
a !-- -. .V.._..A__.l.-i..
-:!-i
-!
- 4J., ,, ' ý.-Li.-Lý.-L
..
-,
ýi
_.
; -1
! "__
. I
..Ii ;!
' ' _ý_I 1-,:_-ý-_l
.
.
-ý6
;
:
Ii 4:
;. -t ; ;I : ;_ý...-J-ý._
ý iL_'__ý_'_ý_'-]__ý
:-.ýý_+
! i .!..!.,!
.
!
;
:
I
;
.
i
..i
.
.
;
.
.
!
..
.
1
:
.
..
:ý ! .
. - '; '
.
.1 ! ,ý 1 i: : :
: ;
ý
, . _ý I i ..
,
. : i I : : k .I .
ý.
.. '; ; ;.
_ :.7 ý'ý.
.
..
;
. Iý m
:
..
L
4
ýJý.ý-_%-;-L,._
....
.
ý,
-...
....
ý
..
.
I;
'I
I
; i-- I .: 9 1!-It
. _.
: _jJm
ý._ý..i
. I _;
.-1..ý..-.
.4_...;..- . - f- -ý4-H -r4- - - i FIt .
__ý-ýý__L_.
-.. ' 1[Li.,
ý.., - ýi . : 1 ' ' -,
-t...
:'
I . i . . ý_ :_ý-----. ! . %a . :::ý
1
! . ; . .- .
.
. ! ! S I : aj . . . . ; I ý . .. : . .; I; I. . .
I
..
- J, ý
.
; ,:
ý ;. . . .
: _ý._'_;_'
: ý'
I:
.
IJ-4-ývý.i --T+'-"4
14 .. r:-; t 4- " ý-!-!_.. - - Vt
-1 -1 - : . _1__-'.. --;ý-'
I
I : - i ;I1.
.: ! . - _'._
I: _.ýý11-1 ;i Ir ..
.-- - - _
;
1
i - .1.. :::- ý_ J-P
I
!
;
- -.1.
I
.!_
---..
_4
-,
_i._ý_
i
.
I
i
, ..; .
,
; ...'
.
1ý I !i ; .; ;
.; ;
- ;_
v .;
;
I : ! . : -- i
. . _7-ý
ý - ; ;-IT
a-7
ý! .
. I r ma
i.
. _'.
: II
; .., ,
.
; .-t__l_!...!.
. ; !..
. : .I ; : I . ': 1-7-7
... ,-.__ý.
; ý :__.__ý_'_.__LL
;
_-1
4J-i.. .
-,
..--.-.,
I
--ý_ 4-ý-- ý4,
I : ::.:.
I .7
ý-..-,__ý
;_1
-I
ý!!!!
-4ý.J.
, :-.:.-.I-.
; . -...
..ý_..._._
.. .4
:I1.:..ý-.,...-..--.
, Ii_:_
ý.4
_., 1
_ý ' __
r.
;--:....
ý.'
_.;
...
;_-..
-..
..
L_;_--..-.
-:I;-!.-.
. ý__
...;i-L-7..
.:
I ._ý_.
V._W_ý. ,I..
;..m._
_......
ý __I-i........
_-''1.L..-t_!ý;_L;
_ý
,
.
.
__ý_ý_.'_ý-.ý___:!-ý.4'
, !..-__
-i=
-i-L.- T."
.
-.
.
.
.
1.
4ý__i
..
....
mý".ý..i..ý_ý_ý_;ýz..i____-_
;-I
.
.
,
.
-,
-.
..
I.
.
:
:
.
.
:
.
'
i
i
:
,
I
,
:
I
I
.
...
.1.
ý
-.
,:.!
j..L
_ý.
.....
...
_--.ý_i
,
t-_-.-r'
.
-i_ý_-'
.
.=
.....
ý
.
't
I
*
':
-7
.
.
__l__L7.-:
.
;`
;
4.__ý_:._
.I..
.....
P.,-13
-i_._ý..4..;_-I..ý.-:11:1747--_
.
'
ý..'
..
...
...
!._ý__ý_z_._ý___.ýý_
_-!...
.
.
i
.:,
-_:-_ý
_=
_'ý.:
..
;
..
_;.;
.
I .; -__
..i. . ' . - , ý..
. ý...
.: ''1; ;I
; . , :.
. I. :..
.
i
::,
.i.;- .._.:_...._1-4.;_; . ...̀i__t-ir_L
ý!ý.': 1'.:4; '..
-.
1-...ý.i.
I..-.
r._iý'_L:_4
-A.-I
_.-11:
-,-..
:3
i",..
.
1m ..-'.
;_-...r.
I_:
...'.-ý.+...._
i__..._._. ___- ý...
.:1.1 -_ý.
-.;.:...
;
-. ...
_I...
...1;
; 1
_.__.!_____ý.:'..A._...j
I-__L...__:_.Lý.;-.:.'.
fj1..--'.-:.
.......
ý...
I-_ý.
I.i-;.
_t
.fIL_.,
L'-. - _..
-a
!-. r-7-ý--.-..:
!.....
;...
..
;ý.-...
, .; .:-. _1.1.':'.
...
,.ý`
!.'ý,
....
I..
i...
_-1---,:!
----f - .,. -ý_q
. ý--......
. :..;..;.l
1-ý.-ý
t..'__, ;-4;
ýý4.ý_'_'-1-.'9.
_"._.
. .....
"ý.__:
!: ; 1--l . .L.J...'
Ji_ý
.._.:
.......
..
.:.:
:;_...,-ý
1, ...
_ý-_--.-.:.... -:t!:_: ý7 !L
.
-i-' q
...
.t_ii j,ý-i_ý._ý_.ý._:..i._:
;- ..
!_L...i..
ý ,ýI, Iý,I! .i":ý4_;...i._!.._
;.1;.......
...ý
;to
;1.:.Fi..i.4_4;. t-..t.-4-iI-I.--;I'.._ý
. ,.:
j_-_.....-..; --',-:.- I. ....I L..i ..%I,.' I ..;
. . . I. .,. .. . .1
--- -!-,.. .
... I. m
:i
.i- i-.;.
I
'...
_.
....-..
..1.'4-!.
4-'.!.__ý.'
_--- .- .
..;
.'I..,I..I-:j;!:1 ;.
.ii..!..
+..i
;.I .t
__6
-! ,4I4;,'..
..-;.
.1t-:.ý
--ri--,!,ý.-7aI ...I
..:;-I.ýi
;I1_;._
'. _i
I......]I.
:. . -'.
!,.ý...'ý-ý...
:ý-ý,11.ý
;
I!1_
.ý
I._ I.
..........
;.
. 44-it.,
iT
'._4
..
.....ýji:ý.ý
...
ý:.
ý___-ý-.:-.r_.___._l_;__..
41!
jý-Li Li'_.:
I.L-LU
...
I-Iiici.__.H..
-ý7........
I -...- ..l.ý.
......
I
' .
. ___ : ý_"_
I
c ý.ý..ýj
1
:.: :.,:.,
. I.
. .ý. .. .: :. ....
;I ,. .. . .j .. : 1 :
. .
....
.........
.....
....... ..... ... ý'_ - ....
. - . . .I . .
. ; - .. . . .
.
.
.
. . .
; . : )
..
... .
.
i . . .. I. . . : i ; !,.:...
r. ...: - -:.
! ,- .:
.
- .:.... . .
.I . . I. . . ..
_'
L..i
. .
; ! .: 1 I ..;.1
: 1-
ý: :
, . I
.
.
Z1 F
1:
: . ; i I : ; ý ; . - II .
: :,
; ý !d -I1.,
: ! -7I
;
! I : t.
- ;i :..
. ý. ýiý I'.
; ! .m. 1
. i1I, .'i
:
: ! .. ,i . :j %.I I ; .. .. .! 1 .f -.
.; i ;I'. : .' : ý ý ,, :; ii , ; : a1 .
m
,,. : .
-It4 ; I. . .I. .. .
I.. .. . . .;
.
. . .
.
. .
.. . . .. - . . a . . I . . .
.
.
ý:
.
.
..
.
:
..
m
.. . . .. ...
. . . ...
. li. ý
.. .. .
.
.
. . : : . . :. i
! i .. . . . . I . .. . %
.
. ..
. .. ... ..
: .. . ... .
: : . . .L . . . . . : . : t :
. .
. ..I- ". .....I, ...__ ,-.. .6 .. .: , I. . - .1
:; ;:
............ ....
-- .
.
,.'- z r i - ......
.
: ;.: '.
.. . :i.. .
, .-- .... ........j.......
:
I.. .
.
.
,
.!
.
:
,
r..........
.....
...
. ...I .
!ý
.......
. : i-.b
1-ý
.
. %;.- : ; _ _ . ; -;.,
....
;
--: ;..! '- ---.. . .I..
..;.)
.
I
,
:
.,
1
.:
;
.
.
p
. ....:
...
...;
!
ý
,
:
,
E
1
;
i
,
:
,:.;! ;!;..---; , :..i.
- .I-
. I. ; .i -ý :1 F
.I
, : .,
. % ; .;
.
I ;I
. . ,It
: .
..
.' I '
r
- "I
a
0 1"
; .. :. .
I; ; .
- : ... . II .. . .. ..W :..
L . ..I ;
.
.
.
.
.
t ..
..'
I
'
':.:!A W
;' :. ý.
. 'i
-;
I;i,,ý;::--..
. !!ý' 4I1..,,ý..!.';.;
1
-ý .
. -;-t -.-ý-I.I II I: .1
I_ý...
L.1
-H .- -r-, .'_iý_i r-, _ý.ý....
1_11_ý-Jr..Iý'_!_-ý_
I .1
. ..-. .+__
.1]
' 'T'
! 1.
i - '- ý.t, " 1- -1O t 1-: - z.1
I . .;ý .I.LiJ.. z.., ......
.. - - ýI i. I~ -. -i- -- . , I-L-14
. --;-., -7
ý ; Ii ý ;:fl.;!!;
! I -: I: , ;
i; !ý f 1! ; I
.!i
I.
;
.. i -'. :" ý! .-' . .-:
..,
;
;
I
;
1
,
.
f-'-ýý-i4
ý"-f'-I
ý
7
-.
.
.
.
.
i
_ý.-i
.
ý
i
I
_. .!__.:_l-'.-I1, :I :.ý1.. - .Z'
.
I
!
,
4L
_ý
......
ýý.
_Li_;_!_j'
ý
-.6
i._I.i.
_.ý
i..i.f.
.
,ýi :! . .
.
.
.
I
I
m
j r"
I I
!.
.
.
; .' .qi ,
; f - . . . . . . ii - I ; i
I ! .; : !- ; - :
-' ý L ' . - --- ! - - i, ,
. q!
-' ý! :
. .
a
.
- 1. i
I . . ý.
..I
I i, ^
. I!
: %-.
I I 1,. .;
i:
I
!i
I
I -.I :I I.--I ýý I
1 :
; . 1 . : 1 : ; i .,
I ; . j .!! .i!I im1I :I i . .;
.
:
;
ý
;
1
J
'
I
ý
!
i
.
;
,
.
I
I
,
.
.
,
.
:
.
Q
!
I..
i
.
.
I
I
:
.
:
i
I
;
a
;
;
I
1. .: . I
..
i.
.
.,., .:I ! i. ,.
4-ý4.ý.4ý
ý-4.i.-ýr7_111
.ý.;...:'__';ý..
---ý__ý_-_ _'.-ý
...'- ...-ýIj- -A-.- ý-.........
.--.
'__, - .-II- -i
-.. -!-.; ; .;-, I"-it .;
.- - -.."Z.- - _i
.
......
:_:ý
-.-"..-ý..
.....
:'.;L"._ý
_..+ýý..;_.'.
12--'.
ý-+
.4
-,-!
-a'
J,
4L...
ý
I
i
..
i ;-4I--. l..#! 1-ý_L.!._
...-.-..--i-I . I.::_-,,
:
-1.
-- _..: r--ý-- '
-4 -1
......
..
.ý.rý : :
it-.-!- . t L .; !__; ;_. . ..4______.
---- ý-1--f.-;-- - -i .-! 'r.- --i; -;ý'i- -H_ý..
--ý-ý-..,. . . -.---ý-..-.ý_r
.. . ý..- .- -.' :ýq ý_I ý : O oll
- ---ti. . --. . ý- ;'7 ý '- T
')..
+ý . .. ..
- -_'.
I ; a . -_L. ..-.-.
. I.ý__-..:
,.:'
L......-_
. .+-ý--;--(-!_ý
; , .!_.ýý
. vi..ý7-:_
__. I .'.'
-:-!.:,#'
!--;.,..j-ý.-;.,..41ý-.,...-.,.ý-.;
__I_-_..
!.__'_-ýý
- . ._.......'-_;_-_.'
I-. I-Il.t
-_ý-j':-.,'--_11.1__ý_!'_
r:_il=
-:-:
.m _
;.1 I...
. ..
i _:._;..j!!!...j:j
I.-Lý..;.__
I ;..'ý
- . "I., ."
:;-L -4
--'-"ý ý-----ýý.11
_;- --i- . -",.
'__. --;- .
'- ...li-r
I: . ; ý. _._....:.-..!..__...
.
.-t..
- __
.!_n.','L::--t'
.L.1
!.....
. - -'ý.-- ...
.. _''.i- ý_ý _-J-4-...
. ý-_
ýý_Lý__
-__
---.
.._ ....
. ' ±.';
_!__';:ý__ý_.-f4
.1.
...
.-...
I
_'
.:
-i..
.
_;_
:
..
ý.
..
A..4
ý
,
.-r-...
!--;-..-.
_.......
.. ..
_ý i-I -4- -4-!-ý
' lý. - 1 -j--+-.- - - ._x
.
-i
..+ !ý--_:
..
. 7.
.
.. -!---ý-i +ý
' 1=1ý
1ý;--;
.L
..:____jl__.L.4
A--,--. -ý
.. !m-:?'ý
' : .mt - -:-,
._!-_4ý-4-4im-f:
- 1I
i.ýi. -!. . ..;__;_ý_z
1_4.;
1.1_
.1
__
.
ý
m
4
.
_;_1
___
i
.....
.
-V_Li_.-_ý
;
'i; ;;_.,-..
1.
---.
; 1_;--,: .L:
_
.
i
..
I
.
c
-_
.
;
'11
.
i-m
a . . -!-.; ý _Lý--. ,.
..;. ., . 14--ý--i"-4-ý-;
,
.
i.
.
.'_ý_-_'_v.
ýý
_;
.
I
-ý--;--.;--ý-I-,
-,.'. ' ;-,.ý'_
.
1ýr ý --..""
, -1. _ . - ;..i-; 'ý_ý_ý_"_i_;_ý.._t
-41.ýý- -.- -! ý, :'
--- -I- -4 ; '= 07
-..
_.
_..ýt-':' 1-- :1---ý....-L?'. .. _+zmI
' _-7
. -7:'--1ý_....._,__.::;
'.i ý "t
.r- -+-:.-----
..I
- --!. -I. .
-r.. --ý +
i..
Im
.1
q.:
tý I I, ...1 -4-44- ý
-i-tt_'ý7.!_4_11
i -:j: _L'_ý
Ii.-_ý"i_'_ý_
1ý_
'_ý
G. 1LII_'_'__
...a.....
. .."-..;-_ý__._
: ..-;.-"
;..--1
I...4_1
. :--_.l_--_;_..ý
.. ..
.;, .itr4.
-.
L
.-.-H
-+Ir
..?:.....
-1T'..",
-;z.-.:
-I.1.
.m..
_ýý_'_;..'_.'
AJ__'_S:...-..
..
....
...
.....
..
....
4..l..._.m__'
_:
-I.!.ý
. V_
..'-.'..
I...11
.;,....-.ý-i.i-.ý..ý--ý4.-ý.ý-!---;
j, .,;.".
. ,--..;-.;fIL
. ý....!ELI
1f4
.-1
,:..
.:_:
..-). ý._
i--._-.-'....
'I--"
...
Iý'_
-Tt .;.'i.-..':`::
'-,._.......
.4.4. .t;L;-.-.
!T.i:IC
i -_"'
-!ý
-1.
..-
-L
-
-:1
.
-r...
-;ý::
..
ý:7-.: -
.;-..
-;-FE ý ·-;·Li.p
--.
:-i .
Z·-, j-1.= .
'
. -- 4-z-,.;-.I
. ----i--f,-ý-;: .
- -
-. -.-- ý.;ý--ý
!f-;t !- -".4- .. -L;
.
I
fi
!. ,.
H-,----;.
.
ý.-
. 4
I- -- ;-.,I
-
t
J , -
-64-.-
--
1
j f"
I i-j---L
-- - 4-,..-!--..
i!- -..:-L.- - ----ý--l_ . : .,!-!
-i
-,
..
:.!
1 .
·--
. .
-- ý
-
T --i
--
. *
ý
I11,,v...
I
)
I
i
ý----, .-..- ,
-r
-iiiijii;
ir
ý .+
- -i..- -- -.- ý I
, ý. .
.ý
-' ----! L
:m- ---i
-: - 4-ý-l
1 - - - . .i..-
:.: . .
0
I
.I.;
i
!
i
I; .
..
I.
-
ý
. I
.-I 1 - -ý: I:
.. I
.
1: 1-:
-;- . -L.. I i-"ýI ..-.
. 11a;,- ý.-!. j .l.-.
__
I
-,-- ; .-'....
&I
'-' -i-A. ::
.
.. ; . .
I....:
-ý.-'ý_--_;
.' ..i!, .I
-..I .-,II.1.4.: I
i1
ý-;- -:+
+ 4-!.-
iI
ý -1I i
-ý- -,.-
i
.
I:---_
..
-. ;..;
I-IIf
.1
ljz ,..
I . . . .Au,
...)
.....-.I,!--io--,
.
i ,I.. . .. ..
.iýý! :1."; ý0 -k
. . ..0. . : ý:
.
...ii,..,I
--
... -ý-.I 1i..i !
iI
--T.ýili.
I..
i,
:
!;
I i- .
I
I
I;. Iý
.. ,-I..i!I!
.- --' ý....
.i
..-.. .;i I
ii
Fi,
- .t-..-..j.1-.1i-·
.%
.i·t·- Q
..- ....0... ..,.-,
eR
(
. I ,
I...4-1.
.: -.
..IL
I-L.i...-j
m..af---- a.L.4.
.1)
-:-_
-i-.-...1
. :iý;-.,..,----;.!t
4-III
...
,
I
-1
.v!
ý
.
ýH
II.....
-7II.7-1...
--a
-_4
-,
..
I--;-t'
-..7...I--n1.
rý,.
.,-ýl
-.J,
..1."
jI.-l;-,-i.,,
PE
-.:.-M-,!
--ý.-I..1
:ý:. !.,I..j,-F.L"
I-,1II1, ....
...
;.:...
LLJ-1-.:
+-I
-..
-::
.!:
I.
;
I
:
.!
I4-.ý-..... ..,
I.-1,..I..., ?.., .,I ..;
- -p
.:-ý
,....ý
4 -,.....1;I.I
.
ý
.
.-.
.I...
="...-iý.
....
,
ý-----L
.
.;
--...
-+-,--;,
;----..
..I-,.i:........
L
.
--!ý:;
-4.: .-..!.--7-1 I..-..
...
-14.
ýl-t:.Lq-L,
..,-I1,ý-:-ýý.-j..
j.
l..;-,-,
.....
..,-,.z
-i-ý
-i'm
-..---,
--;:
,
,
;
j.;.ýN
-:-iw+ý.r....:.
:
-L,
.:
+...
1 .II~--....
:.;11
...
ý......
..
J-L
-;.-;-;-:
;i-:
;I.Iý,-;.:
ýI-...i_;_
.,I
:-ý
.L
4ýt.A~
.I..1
ll, :.ýý
-i
.,----i........-i
i-.w..
"--..;..-r-r7
.
..
:.
-.-7-.:-+---lt-,
ý
ý;
7ý-1ý7:
-I.='-7ý:
!
.1.
.:
...
ý'.;-.
Itý-;
;.ýý.
. ....
.-..,-I----1.i-;.
-.:..;
.-.
:......
.,;--;
I:-;
.. i....-:.:
, !.ý
..,-.
I
I
:
-ý7ý+ýý4ýý",
-;.ý,
:
!-t,
.
, ----ii-:,
.
...:
;. !i.I,ý
:-i.-1.......?
:I.. :....L.:
.......
ý.....:...-..-..-.i-.,.:..4..;-:.ý
1
...
1i...
;
.
1.:
m
ý
-!..ý
-ý
ý
..III
,:-7-r:,-;,. :-. ý.4-,
. I.. --.
.:
.!.
.
;. --.---ý.I-..-.....
-1.ý......
.!
;.,i-I...I
.I .:..
.;.-;
:ýA.:j.
!..7:;ý...
..: ...
T
:..I
.I:.:,-..,.;
:
r
.
.
:
:
I
-.ý-,..
.......
.....
;... . . .- L.:.:.
Lý
:.!ýý.ýII:..
.I.i ..:
-.. ...
ý.;
..;,I
..-L
,
.
-.
1
.
.i
.!
-,.. i!:i:
.1..
.m. I.
.-...
l
:
ý
.
.,
,--,
;..
-..
.4.
. I . ;;. .:.;.- ::iý
;i
;. ; 1 7. .
,
+I
.I-!..Z--ý-~~l..:.
i! : .
i
.
zj I !
iI
I
.
I
!
;
. .. .:. ..,.
. . -,.,
.2
.. i-;- . T
, : .; . %
.::
1
.
... 1I - . ýd.:.-:.. .-.i,.
I - - -ý
I-1:
. . ."
. ::-- -, ?,r1,
.:
i, : :. i : .; t ý: . .
- .ý ....- . .
. :, : ,
m i : ! ;;
I ; ..
. .;
. -:: .
I
..
.
. -:.-.--,-.!.:--..
"-!.......
ý..:
!I
. ; ;! ; m
.,
-.....
ý--;- :
- ;....
.,.-'
-1 -- !..'..;..ý_'...'_
q
!:ýZ-!:;:!;.ý-,
.
.
.
l
t i li
: .: ;
' ' i-
I
j
I
I
-
.
. ..._ .
:
.
. ;-
II
I
-.
, ...
. 1 ....
. iiI:
.
a
--
?
:
I
.
... ;
I
ý
I..i ;;.f L;...
I . ;._.i : r.1 -j
: , ;-
. ý
. .
. .
:
.;.
-I;.
.i-...;+.:--
"ýr; -p -- 1. .
- -41 ý
ý
..
i i
-
: !I
I
I
..
.
-'j --r ] ,7
- ..--.-. I--,
;..
, .. .
-
4z
, . .,
--L. 4,-
-.
---
ý--- ý ,
ý- -t-- -r
.
,
M m
7
-.. ...
. ...ii-..!....
I :.i.-ý
- - L-
. - Ii I. i j ; li
. ..i ý. i. . ii : -i ..
.i ;
.
.:
Iýi , :
- -
ý.
-; ; ...:...; ..--&- 4 ý ;-., - .;..
.41
I
4-.- -.ý-!...
.. ý.;.:.: :.11--i
CR ..:i----.---.ý-=---r-"t--,--,.- - - . .
, ý=:1 i,-J.1.
i.
i
ý
-,
..
I
, :
..4.... I- - -.-.-...
.I .. ý-!
I,m
- I . . .. 1mm
-144 -:--L- ,....--I- --. ;:-1;- . ;- .!-. - - I
; r ;- .. .
. , 1 ., i ý
: i: J- -
-..!. II 1.. -. --...--.
.. -.....ýLý:4.ý.4 -!-. "1I-- -.-.- ;---;-..,A.,
..f..--.
L.
I
II
-P
:.. -ir-i-.
..
. .
--
.
. -
L
-L-
. 4--,; I-jý-. .- - ý4 '.-:
1
.. : I
"r-;--
1
ý - ----
-;
-.-.-.
ýLZ:LmL`
i--':.ý-ý,ý".T.,.,,:
I......
;-:
1ýý
-tý
. j- &
---, 1.` -,
..P...
..'__.
.. I-- ;-4.-T
i .;: I . , -,. : ,-. i-I
_I -4
...
...
:...
..;.-I
,: .;--;. .--, ,-!'.-1.
.ý- -ý
t -.;c.... .--i ...........
ý -; !----,-'..
...
ý...-i
---i-,
, I
-ý-r--..
ý I. .. . .
----. . !I
-..--i- .1,- ... ... L
. . .,...-..-.p
- .- ----;--ý .- 1 -- ý-ý.-...-..I-T.11 : 1 , - - ý _. . :
. i;--iler.
,
I,.
- .
:I :- .r...
,-,.;-ý - 4 i..:.- ..:........
ý7
- .14
- . -.. . . . I . .
.-....*
:
-:.,= ..ý-ý ,, I
i. -H-ý-i.
- ; t." :.-!i r. ý-1-4
I-1-- Li -:...
ý..;..-:.-,
..; . ' .I,ý-:.I-.. 4-L.- ý-:...,- -,
:eý -ili
. --.
. 1 --""-'- , - -,-. -.
- ... , . . '.
. . m"
'i . I
ýý
--. "4
.1
.. W.
...I ! .,.ý.-.-111.1.;
- ...
- i-ý..: I-..;..ý.
p 4
; A'17--l ,iij.,
.!. , .;.
, :..;.
. I- .ý.,
--... ...----,
. .-:ttý, .
.,
- ý; .I .;. ,.; ,'Lý. ",..H -lw zl
I , ---.-- -1
:ý-., ý.
- . . ý _. L
1
.ý........
....
: .: i --ý--....;
..--- - .;-ý : lu
. - ._j-.
;...
.T
-ý-,-4. ' ;-----L.-;.4.q.'
. I , .i-;_
- ; -:....
= = , ,
~i..
_
,4-
. .i I .-..
, ;m
- a.. - . .L
. .
--.- -
--
. I.
-ý
. -r :
-1
. .'.:.
c,--.ýý
. ..
.,--... .... .r- .ý-,-- 'ý..
i
ý--
L
.-
-
I,
.;..4
--------;- . . I !-C, --i-,'-ý-J:441
ýI-L4.,
-,;--.-jE
ý
.
M
.
..
..
.-..
-:P .
4
-T
,
-,;-ý-r
I
.
-1
.
-- J-.ý4
-- ·- i.i -.i-- ;--.
+ýi.4.,.-.
.-......
I - *Vc..~-.---"-1.
-: r-s..,w-..;
.......
.
1.4
. I- ýI.:......
ý;I - .. I,j ;. .%.:i i ; tI, ,.I :tt !1I!I 1II
%
-1-I i iI
. . ý.:. - ;. ý.-:..
-., I!ý
:;,:tý;;:;j i,. ý
;
ý,
i Iii
I:, . I. ... . ;t i :
W
7-.--!1-1- 1'..L.-
. ....
F'
..+--q-L.
ýý:
^-,7-,ýý---l-.;
- - -......
- -.... . - -. . ý..-. . 1 -,
-I..,..
i -ý- -!--%
; -ýý.!-.!-ý-i-i.-.i-ý-I........,...Z..f-t- .- --..-.4:
..
" +`.-4..;- ...
7=-.',-i-. .-I
:ý-4-i-ý
ý,--.-+ý
ý-.,--I
!=
-;. t..-:
--t..-...
---- - ".1--i7.
-7-..-.
-.ý..
- --,--. ý -ý . - -i- .i.-,..
; -.. I7-, 7M,ý.+.7. -.. - 1 -A tý-+
,ý-,,j
: - ,-t
-I
.--.,A-,!
. ..-; . ....
.1-1-4-i---;-,-ý
-..,.,..
; ';-7'
ý;-,L-..;..ý.;.-.-ý-4-ý4,.ý.ý-.:-j-' ..ý--ý.:
tz-.-=-,.
11,
---ý.,I.
- 4..,.---,.-z!
...
:,.-..-..., ,--..,.ý..-I-ý
-44--........
;-4---ý,,....--.
-17.----.ý
....
-.. .., ..- ., ý -.-..-.
%.:.
ý.:
ý- ......
.........
14--m
,--.
ý.. .. .. : . ...
.
-..
. I ....ý ....
-4--,-:----,i
j.;-.-,-.I.,-I-.-.-.
ý:.ý.1,.. . ...
--... . ,;.I
.
ýý
..-- ,"....-ý.
-m
.-.1[--,;.
!-,.
I
'.
.
ý
;I
.;
.
:...
..
.
i-ý---.t.ý.-ý.
ý----.--.-....
l-.:
.
.
i
..
.
.
.
I
.
.
.
.
__-,
I
.
- :..-..
- ..-.
.., ---, 4.
f-ý.-:-.:-'ý- ý-4--ý.-ý-,----! ....
It~-...--q-,..
--....:_...!'
r-1--..
I. I.-. ,,-ý :1. ..-mI I.:-....
, , = :- L-3
.
..- .!--,..
.-. ......---. -ý-ý-i-.m...;
-..
: ý.--i ý -...-;
. ,. -..
. . ,
;-:-..-,..i..
ý4 ..- .aý-4ý-ý----4-.
...
...
ý,----ý ýý.
-l--4--:.-I---.v..ý.
.- - v1.
:.
.---ý:,.ý ;f.-.-; .7 "'
ý - v--.-ý-ý-!--. %
ýi--.
-ý-.:d"-. .-qT-.-. I , -
.!
-
- -i.
.
r."
-,----H,
.
---
.
-,-.
. ý.- ýý -.--.
;ý.-,-.-. .1-1
.-,- -,:-.ýý.
-.7-
--i.-:,,:.
,.t,-.-zý,ý,.
-'iý..'.
.....
f.- --.
ý-ý.,'_I't 7 ..'I.- 1-7! ' i --I.I...
ým
.
i -:.-:.2-.---L,
...
.
: . : . t!
..;- ;,
- ýý
-- --- -...,..;..;.,_,- 11 ý-t,.J
ý.-ý-,---,"-ý.,4
4-.. ý-7n,
. -- 7r. ý
-----... , ,
- - --.--,.--i.--- Iý--.---,-ý....
. 4---r
I .
. ::
... ..-.- t-7,. ý . . . .. -i 1-.. ý..
I- .1 .. ...
ý-.J.., -1
r...i.
L-1.
..-- .w. ----.-..-...;,... -.-ý-..-,...,....-.:... ý..-. .
;..-ý-;.-,-+......
t....-,.
. ý
.. . -t--'--,-A.
..-.
.
. I
.
.1...,..
l- . i ý,-.. ,- - ....-.
F
.-ý
. ..
,
ý,--,__..j-.ý-L..---'..
.ý
0;
.
0
-..--.--....... .
- .
-- - -1,;:,.." : l . %..:......'.
I 1-i - ...: L
.- f-.1..
t:ýý--: -..-",:.:.
.
.
..
.
. ý..
.
.
. . ,,.ý_:__
.
.
.,..-:.J....-I.----.--I....,..
...
!_,.
-:--,.4-;ýý
ý i-ý
- .----!...,.-;.--.ý..,-,-;-.ý-ý-i.. .;.4--ý- I--j-ý-;-. -4--ý- ý-41".,.-'..! -j.- - -.. i-I.A. I- .04
.;.-ýa - -L . ...--...
-. -ý--.ý.;..-.--..,. , ,.. .
..;..
.--;..
i.:..:..-..
.
.
..ý
...
..
F
,
.
i
!
-i
-t
.
.
..
.
..
-"..
L
T
,
..
....
ý ý,- - : S , ý .. L-.: ...
-.7..-..-.---.- -.. - .
.- ý--I.I.;-- ý-..,. . ý; I-ý, ..--.- A...;-I,-]-",
: - ---. .. .--I . ..: ...
t
.
.
.
.;
-.:-i1ý.._;
;_ý-i-,ý-.
;.-;I-.i
ý
,
i
;
-i-%
:
.
:
I
ý
I
,
.
.
.
.
.
l
:
.. I. -4 ý .
- . - -.--l-4...; .
I a
-.
. .-ý i. - .. -.--.I . .. - .!.
-.'! '.
-- ; -...ý ý.'
_ -- jýA-J-..,-ýý_ý--.:
ý
'L
-1--...i.ý-l
;-.
.
.
.
1-4..
.;.,.:.;
,
....
- .. -: j,-'. ;.-..-.
-- I ..:-.:.
. t.I i
:
.
.
-L .- -ý-1-7
, ,
-... ;
...--,...-i,--....: -L-...J..
.... ý.I.
.....- .- .-I....
--. . ....
..
....
I7ý
... . -,
..t -:-,
I.-- .;..-.i
I........-.
, ..,
- - -ý.:,ý-`.. . . .... .. : . ...i,-.I- ý...,.:.i--.-.:. - .--.
,--::; - -jý -.-- - -+- --- -I ......
,.-- .-...---:..:--.-- - -.. .......--..--....-..-... ..:: =.+:ý
-- ,-.-.
ý ..
::
. -,='..
- -;'.'..-ý: ......
ý
:
. : ..- -: --:.-.!,L, :. - .ý ;..
-:T :iý-,L_ý
I ý.
. ,;=.:
.:. - ,. : -ý ý - ,-ýý
..
...
. .....
. 1. . .I .. -.. ... ..-t .:ýi--;-.4-:ý...;-!I,
ý-4... -r ;..- , . . - : . .
I.:-a l--I-...
`i:-,-i
ý`-,.:i4f
.I
I
.-I
q
ý
..
.1
-.--.
A.
--...
,
I
.
i
-.
--/714...:
r-i-I
L...jý-!.- --ý----4
. -i.).;.--;.,ý,-:,,ý
... ..;
. I--.--ý.--.1--...
-ý--+, f-4.,.4--ý-ý-H-:..
.:
jI
.:.!i-l.-.
I-;--,!:--ý,.-,:,,,--.ý_-_ý-_
.Iý:..
tiý..-I
. Aý
:m
4-..ý..ý...:...,I.i...ý.-ýf'.-,f....
-ý
.ýr...
...:
--!-ý,
,ý.
.1
ý ,,ý,
.",--ý..ý.
ý".,-a-4---..
-ý
-ýAl
..;..,----ý.;..
---f- --,
- -ý..
..-...::-..rr-I
ý,-,-JL..,j.
....
. --,-...- :ý-.
Ill
-,.!
.
."..
..
.m..
..
.I
i
--ý
, .....
-t:.M.
,ýý
L;.--.-.--...,..-'..i.. ..;...
.--i..-ý.....:..
l.-7--I..:
-...
..-.
;.....
;...
4'-r .;-ý4i.-.
"T-4ý!-..-,f'.--7-.,:I!ý--ý-.-ýý-4---,Lý
...
-7!-.-...I..,.--.--;
I
:
.!.
.--1
.1
1
..
...-,.-.
-:.-I
..Li.--:-ý.ý
.i.
ýý1--.
-ýý
-ý
.
.
4.
.....
,
I
:
;
...
,
C
ý
!
.t
...ý,.
.-1.......%
...
..
.:.-.-..ý-.!
,
,
;
ý
ý.ý-.I----.-ý..: ..ý...
I
-..,
,.
--!.,,-..
.
:
.
I
!
..
.:
1
:
:
a
.
,-ý
..
ý.,_;,-j.
..
':Lý
;lb
.,-.-:.
;ý.
,-,-.-,--.--..1.
.ý
.-..ý
,.:!.-:..-...
...
..ý.....J..I..;--.
ý......
-.ý
1
i:.L
,..j.
.!].',.ý
;.
.
.
w....
.
.
-.14
--,....
ý,.-.-i
--:.ýZ
;.!
.......
1
-S.
-.
4
....7..-..
..
,
.;.i
.
--!
-:...
m
-a
-.,.!.:...
-:
....p.:.,
-,-----".
.-,iýI.,
..-,
:I..I.
,I
.-..
...--ý:i
,
1
:
-ý
I
.
-,
I..
ý-ý.
.-.
--,
.
w.
f
!
i
,
.,I
.
..;
-,-,
W.-..-,
..1
:
::-.:--"..
i-.!.
ýý
;
No
-I
4ý---i-4-,
,
,
- I-x ,
1
ý
.j......
!
-tý,-..
.t.
..ý
i
.1
a
..
i:
.
I
7
..
.1
---,.:.- ;
---..--ý-.--!
.ý:7
..ý .1-.
-- ....-I 1,
-,.--,- ".
"" - .L.;
+-i------.
- .- -.ý--;-ý-----ý"
,. I--1-ý'ýý;:;.a-ý.ý
.;:::i.-..---j-...--.-......
I..--1
ý,a!ý-:L---;
....
-......
-:.:;---,--.,4-:
=11..ý..;.
:--:, --ýt-ý.'±=.ýý.a:t---t
:ýý -ý-:Ttt
iI:,-;:
ý:.,:t;
-j,-.ý-.,:.-==
z.-l.-;--,.-,.:
...
-I.-,..:
---.
.,-...:..ý
ý:. -..--.....
E:. 41- :.4..
...:Lý(-,..w_.:iq
ý..!
. -1.4-t..
-.14I------..
..
......
- ý,..;
.., -!.....
., ----- = --- ' ---4q .. -...- 7ý# ý
--- ,-'---1-t-
....-...--r...
t...M....
------, i-a
ý. I--....
ý-ý::L,,.
L-...ý----ý-:ýLý
.- -....- .----I-- :.ý..,,..;-,
1,
.F
--,---,,
,-,---,
I---".--..,.....I..--+
-I-..:-7
ý ..: :. -17-7
.. .
--j. .
I ..
.
. . .
1.
; : 7 ý . T`
- -.
--v
-: -. -. -- -,:.- -.
i ,. . :
ý4 -ýA--.
, 1.
.
-- .:- ...: . - I
:-ý,..ý.L-ý...._;_.iý.-........
2- j.
.rý:ý
ý,:ý:
j-,. -ý-..- t - -I..-.--i-ý
. -- ý.-4
. - -,--ý--;- .::.ýýT---....
ýý- L.....4-.i---m--j-...,
.ý,a...
,:ý,_F_'ri.ýý _ý_ ýý
_j_ý
I
--. ..:, -L4---:-...
m- -., - 1j...
. .
+---.....-.
:
I
.
. -. týý...-A,
..I..
.
-- - . ::-. -.
.
.
.
:
I
I
It . ..- ---. - - . .
.
-.--,-i,
.
,
...i,.!.-.
--ý r..-.-- r-.-..F..-:...
ý-,--i,
. -i"=ýý.ý-j. - . . . . .- . .. 7. _.,__ I . .:..
.-. t..v .:,;ý-;-+----H-tZ..
_: ._I. ` -,,-ý------fZ. .;..;.,-.,
ý..i
-,--.ý;
..1 ;_,
"..;--ý ý-:---!-! -!-.'.-t-f--..-, ,wý---.
4 :,:
. ! .- . ::....-:
i -. :ý::....
-_
-,-,--,
ý-...., I ý .-.
.....
.. Z....
4-.
ý_.;
7' : - ; -1t- - ....
-.- r-, ;
. - -_!
.. _1
. -ý:,.-t --, Z.
. . -.ýý-!- I
ý,.,
- =:-ý.. ý..
I. -i. .; .. .. .I-..
. ::,-,.
:- ....t..-ý.--ý-,.
. -j-4--.
.
-----I- -.-............
. ...
!---- -4-77wý - -..,
.--.,-ý
.
-...
..
-...
:
,
....
I:
7--,
-.
"
,
ý.
.
.
. - -.L.
- -,-- ...--..-.
. .
I
--:-,i -; , ý -4. ;-!-;---- ,--:-f-
. .:
...
.....
..
41Z-:--s. 4.---;
ý..ý-.:ý
...-i-..,1-;..--'
- __-.:
E4
-1.7
7:=
7---"I .-,-j
L-..-...-.I---.
--N-.I.
ý:.:!:..;.-:
;.... ....-,I-I
II....
::ý
, 1ý
,7ý,
-. 1.-zý. -: - ;1 1 - 1--;
. 1. I,-ý-I
:..,-...
ý-;.,.-.-..-,.,ý;;
.-.,..,4 -:.ý-.-.I..:l.
-.-----;-.:--:.-ý..:
..----,-;...-.
1:7-,-,2,
..1 ..-F
ý.
I
r,
.-1
-i.-,.
,.--..-.ý
ýý, I.----...
;
.;.!.!:..-! . I....:..;...
!. ý. ý, : ii,I;: .. --::
,:"-t
.
-"
..-.m"-7 -.-...I
i-4
.ý
ý- - ; 4 -7---,,--.---11.
.
--.
.
.
:!
!
I.
ý..-- - .
.
--.
:
..
..r
..
I . ..:.;I
. :.-i......
...j-ý..:
ý
,i......-1
ý..::
.. --j..- ,
. J..-.:--ý..,
..------- ..-.-L,---...,... . - . . . ---.ý.--ý-j
I. -- ;.;-i-.;..i..
!; ,-,,
;....
..
ý-- - -.: -. r
I
-1-L
,
a
,
ý
:
.....
p
i.:4------------;
.
ý
.,.i::
.
,.;.-.
.
,
l--ýt
.. i-i 4- -i.
-II-..-.
..i.-!-;
F
..
I--!-.1.,..ý..-.+...-i-..:I.;.---:.o
--ý7.-!
-------,.4...-.,
. ...
-:ý:.
7.':--. .....
. I II. ...
:, -.-....
. ,-ý
- -.,..
-:, 7",
.- ,.1.
-ý-4:.-... IL.-t - --,-.-..,....: ..-...
......, .,..---..
-:..,ýý-ý
I-:
.-.-.
.4.-.--.ý-ý.-.
.+.ý--Li.
! i ....
: ,..:..-.
:-ý.-..,.;.;..:.I.-i-..-i.--.ý-;.ý-1.
-...:I....
, .. :.I ; .....
.
. 1I-;...
. . -,-i.',-,-,
ý..
11--- .ý.ý
.:- - -. -I
-- . ,j j,
Lý .. !- I, .-t
-:. .- ....-.- .-. .. ,-7- , --... .. --. ...,-:-:.-,.,,-,
. ; ; ;i....ý-..
. i . I . -1.4-Lý%.
i .-!..-.--'=
m1.
.. . T - z.......
. a.:.
r . . . .. ..: ..-.. i.-. ,ý,
. j: ,:-.
.
---------.I- 4.-- i .. ,
:
-L-i---i.!...I.
;!,.a
-ý_,--..I-;.-ý..ý-.-4-..-,.-..L.
ý
j..ý._
_
.
ý.
ýI.
;
' L.-., a . .. ý,.!-- ... 7 , , I ; .
...-..
' ....-.
....A,
ý
.r
I-i--.--..--.
1
..-.;
;
I
.
. .-.
.r
-I...q.
-1.1-1-1-..i
.-I.I.
-i4'-ý.-ý-ý'-..
--...,',
...-.-....."
---ýý..d-ý.ý.....
.
.....
.,
.
;
J
.
-.
L
I--ý-'--L---]..--.
ý
:..;
.ri .-...--..t.. - - -...
r -;1-...I.,..
-i.
:ý
-ý
. ,I1.,- ý- '"ý
.! 71--ý-1- -i--ý.i
-. . Jý.....
- ý!.-:-----I
I I.:..- - . ý
, - ia: . :-..
1--r4-j. -1a-1--4'--!-- 4 -- -- -:-,
f.--i-.;.
. .- -: 4; ..: ..;Aý"-., ..,.-.4.1.L-.4-1-..ý-.ý.ý
m
wwL-. !
:
ý,..-.. ý-----.-I---?---.j-.,ý.. w-im--.....:
....
.q..: --.
!
:
,I--,
;-t
!
-T-ý
.
-;.: .., .:...
I.
. ..I - .,;-.
.ý
, ...
Ii..- "I---..
. i-, .:-,-, ...-,. ;-. A-.L - ,---..
ý- . -.,---.!.-;-- . -L.- 1-: -ý-;. ,. . . . I i . . . -j .- ;
. -1
. .I .
.1 . . ..--.-..--.
1-4 I. i.. . ..
-.:ý1-ý,...
I " I . I . .. -4- , . ; ý, I.j..ýý.
. . .
- .
m
. . .-F,
-1,.. .k.
It I .
. . ...,..
7 . ý,: -777
: ml..-.. .
. .;..
I .ý. : ;.
I ..
:..t-.-tl-.% .iýr.jý,,
L-...
ý..--.:...-.
:.
%
I
.
I
I
.. 1.
.-: b...
-m-:ý
..
ý
....
.....
-- .----------.
-p:;-=77-..-.-..
-r-!!-.
.,
-!....-..,
---"...
.
.
...
,--i.
-+-I-!
.. -:!-A.
.
.
!
! ý.
L..,..-.:,.;---..-ý-;
.
-...
,..i!....... -ý....
. !;. -i,. i..;..,.
s.-:-.-.-"-..-i1--:
%
, j ;-I,;.-"
..:. ...
,T
. ! . ..4-:ý..ý.;--ý4-ýý-ý_. I. . -,,.-L-L!4-4-1
. -.. ý1-.-.......
p....
Iýý-,ý--.1-11;...J:,-;I.ý
r!....
;H,-;-:+.!-',.
-.
ý.
;
.
L::
I
;
-1
I.i
,..
:
,
1
.
ý
-----,
.
.
'..
-I
,
,
I.,
.1
.
,
1. -1I
,
...
.
.
I
..
'
-l-,
.
'
.
.-..---.,
ýL.ý
.11-1--i-i,
.
mi,
,
I
;.
,
I
.17
".
ý
Lp
.,
:
...
-.
-:
.
. I.
- ..ý--1. .. ; - .. . .j...L.ýý
.
,_..;.4..1.:ý.-..
..:.
U
,
I
....
.
I
-ý.
.
.-ý
..
..
ý.
,
.
;
.r
l
..
;
I
,
.
1
.I.
I -ý.:. . 44 , ....i... i!.., ,. .!7 ,, TIýtý
- ,, :--..
i -!., .1
m.......
.1
. -.--.
. z . i. . 1;. ;I I--: :--, . r I.!I:
! ....- .-:..7-....
. .. 1- - : ... !;..;:m ý. - .- -ý-T
i ;- ;ý,.. -L.ý.f i .,.ý--2
,.
..;..
ý .I .-.4..m...
i
;:
.3
..
i
...
i-.,ý..,,.:
, ..--.
-i
,. . .!..
. .. - mý
.
.
,
I
.
.
.
;
ý
,
,
:
-T
I
.
;
.
.
%
.
t
I
;
.
:
.
I
.
.
.i !
.
...- :.-- ;-..
..
t. . I - ; .: . .
I It;
-1
-4t . ý.-;-ý-.-,,-i--ý-4
.ý-.-..;
... : . I .. :. I.
j -L .
: ý--.
.
..-; .L-L
.- .. -.r....! .
;. i-;.-;..ýý-4-ý.,.,.
i :; )-..
: ý , i.i_-ý.]--ýý .,-t
-,-,
. : .I-.---.ý-:-f i-r--: 1.
I ,-.-..--.ýI-..,j..
,-.
- . H .-I-. I. .
.
-ý..;ýýi-i.i
Iz;m
I .
. ....
.
.
I .
; a;.
I ';
r a: . m
JI1 7.ý
. :. !
I ; I ý.....
. !..
.
ý.... .-... . ý. 1 - ]i...
L1
.
L
_;
..
J,
.-...
t
...
iL
...
ý.
.
. ; ;.
! I
;
.
1
,
;
i
.
:
ý-ý..
...
:
.
.
t
I
I
.
.
t
.
;.
.
;
t
.
.
ý
.
I
.
I
;
I
. , . , --,-" . . .I...,.,
. . . .;
.
i : .:.
! 1r ; . j I . ; I
....:.ý I...- ,-..,.----.-.:..?-+-i
ý.
. JL. ..!--L.'ý...ý..;
,. ý :ý.
".. .4-4
. ;. .ý-ý
,
., _:.i..:.-:-......
.-.
. 1I:, ;, ; .1.. ,-:!..
.;-..-...
. 4-.t-..-... . I
...I c 1
-!.
i+i
. : .ý a ý, -.-. .
. .:--t..i
. . --. J--ýý-_: L. . ; :i
;
....7..I1-. .;:....
%
.
I i .
.
f.-.ýý
ýI -t I..I
-ý.ii
" i -;. -- -ý ..ý-..-I-.-.ý..-A
ý.4...--.ý.,.- .4 -.....
. . !..
-I. ..--..-1 ,11: i ;I II . :. ) ýi l- ý ý !J : !ý ! , !; ;
11llui ;.:ILLI
i: ,i .
, . . ..
::... -ý.
.3
;f
t: I
4.
..- .I.....
;. .,
. ...
: ...:I...t!"
.. :- --j,
:
I
. .;
.
.1 1
; ý t : . .ý I
I
; 7- .- ,
I
,
:
:
I
;
-r
-ý--..
.
a
:
I
I
.
.
.
.
.
I
:
I. .
. d -.
:.-,..j.-L-..!-,.-,
.- ___;
.
- ,I ,
,ý
1:.
ý.I...ý..:--,ý.-,,.-ý
I ý ;. .M. I
.....
. ..
ji
,
.
iý---- ... - ;i -.. ý ý .. - _-;
, t I : I ;--;
ý... - I.
. - - -i ý. ......
;
-,-r.
.
r
.
,,.1
.
;;
1
.,;,.ý_2
.
- - .
,
.. : t : I w i : ; i ! , ý
.
, - I .I..
I
.
.
I
I
;
.
.
;._
.
ý,, I
I . ý. I
ý-, , -..-.- .m:
t :. .. 7 . . . i I : . - ! ýý 1 ;t- 1-, -: ý.- l
:
1 ý---- ý..j-i..-_-L
I
:
.
.. .,....ý
ý..
.
I
.
;
.
.
..i :, . . , ; !.ý.ý :IL l:l ii. :.., .q :-!A
;
s
:
.
jI ....
.
:. ,A
.
ý.
-.
......
-rf,-ý!
I
ý
.
!
- 1 --- : : .-.--. ;
---:i
1 iH: Ii-tý
,, ýi
I!; : -1- i
ýi,
!.,
-.I..
.-I-JI
"ý4
,-4.Ii-i-. ..k-:--- -...- ý - ..4.+-..
ý-,-.i -;-.t - 1 4....--...
-.i -..
:.
;
1. i
--r
,;
.ý
1
...,.:
. .. : . .
i
ý
..:tL;.;
3
1
1
1
1
i
j
:
:
,
i
..
f
1
.
.
.
.
I
,
ý
.
.
.
.I
-,
ý
-1
:
j
..:
..i.
.
.
.
.
c
:
2
.
.
;
:
!
;
;
:
;
!
i
.:., : ; ý ; : . - ; .
..,
.
ý.
.r
-.
--r,
"
---.
ý-.It
-l'..
:
.
-;
-L..
.-ýJ--t--.L--iý
I
,
t
.
.
.
-:::
I:
!
-.
:
,
:.
!-.
4;
;
.
!
I
.
--..
11
.
.
7ý---:-t
.
;
.
:
,
.
.
!
-Z-..-.j-----I !-; ::-i ý!t i. ; , 1; !
; .l .s
t; ,ý : I a. .
1ý;.:I
I ., ! - la iý,!-! 1.,;ý ;.!
1.
,;
;! 11;
,9-.-i-.
;;.,;;.
:.
ý :I
-i-....
ý--t-ir`
..:-ý
l'-!
..I
. -- .;. ,,-ýt!,
I
.:
. -%--:
t -.: -.+
.:...-i
m .T.
;....
- ..;.-;-..
..-- - ",
1-.--ý-.-ý
.. !
. .
: -..
. W i;ýý.! i.:
-"' , ; !
ý-A .:--. ,_.;
: I:.
j-.,;;;:. .
-.?. I-....
-. .,. ý,-.-.
!--_i-ý-!-ý-.,
...
z, .I
..
.-.--L-ý-. .i
.----.--.
. :---,..-...t-.--,..
.!.: . . I. i .i ..
p
;.,;so
i ,. :r..i
L..
..
t.:
I i-;.ý
ý;ý-t-...ý--1.ýi-7.t
. tI
.-;ý.....m
ý, II:,ý.f :!
r., !
.:
. :!.
i f :;'.
:1::..
ý-:.!..
,.
i
.l.
;
.. ., .. . ý. ; . ýý
I
,liýýi-!I!
.
mi .
.
.
.
.
i
..
.!
I
,
;
;
I
.
1
1
:
.
I
;
1
:
.
.
I
I
.
:
:Q
;
.%
.
I
.
I
:
i
.
-r7
4
.
.
-"
.
.
.
I
.:
.
4
.
t
A
4
-.
.
.
, .-.-.ý--.j--...-.:
..
:
.
,-.
-..--.:ý.
.
.
m
..
.
.
.
.
.
-.
---:
.
r
:
.
.
.
.
.
.
:
.
.
.
.
.
i
t
.
i.;...
L
.
...
.
.
w
.
.
.
.
.
!
:.;............
0
. . . .. .; i .qvwo
.. . .. .,m.....: - ..:
:. m
ýý:
.1
. : -.... - ý%. ... .. ......
.
II-. ..... ..
,
.. ý
.i.:...
. ...
I; .. .... . -I;. .:.. ......... .. ....;...... '' ..I
...;;.-.,..
.. , . i 0I .-....
. -:.. . i
......
Is::..I:...
:. .
%..
:
.,
....
.
r
.
.
.
..
.....
.
.
.
.
.
.
:
..
.I..
..
.
.
I
,
.
m
.
..
.
.
I
....
:
:
,
%!
I-f
.
.....
..
....I , .-.i.....- . .. .-.. I- :.,-...........
: . I 1...
--t .;
.;
. .
.
. .
,. ... .
;
. . ,:.
. .. I
, . i. .. I.:. .
, ...
. ..
: .;
.. .. %.....
f
. ..
%
1; :: r, ... . . . -. . . .
.:
.-. .:
.......4.
,
.
. .z
;
.;
I :ýý; > s!ý...
.4. ;;;;:,i
.. .
.
..
: ,I.I
:-,. ...
.I..
.!.-tý....t.
;.
j.
f,-...:
..:
. i..-.-,I ,
1:;.;
o: .
-.1:. Ii:,
ii-1-i- i..-i -ý,44
,-I ,.-1 . .
,:;;
.:-,..--.--ý--.---..
. ,I.. i
... ...:i......
-.
!-..-j-+.. i- ý-ý ...
..-. +.L ý...
.
.
.
I
-z-.,.:
,--.,-.,-..I.--;,-.1
:
...i :.-i..-i-...- !...
.
IL,,J.-.-,,ý
.:.
;..-ý,..
...
ý-j.-:
m
.'
4
:4
_m
- . j - ......L.
!
!
.:
f
ý
-!
ý
.;
.
i : - ýi ! P ;1 1
..
,ý; L :N-,, .,
,
.
i
;
,ý
;
.
;
!
.
:
i
I
1
.ý
L
,
ý
ý
,sc
1
;
.
:
;J
;
f
ý
!
1
.
ý
!i. -Iý.i
ý .. : , !, ; . ;, . .
Im i . ., ..
,.. ,
.: t . 1
,..- . . .. .-I I .,-.1.;..,..
-.
;-! 1-1-r..
1-7
"+,
Iý.-.-+-.-.:..I
-.-.
I
,
. . ...
-ý-ýýI.J-L!+
i,
.
i
;
-.
Ii
1
1-1
..
I
,
I
.ý-ý...,
.
+
,
-!
-..
ý
-ý
-'ý,4
"
:4
..
-;-4
- -4L--'
I,--'-.
I..
ýI
.;.ý.ýi .:
-ý. j.!-..ý,j-j.-ý-----.;!--ý.-ýý'1
I
. .
!.f.....
ý%:-;ý-,'
.1
.ý
-I
....
:
Ij.
,
;ý..j.
;i.l
!1
.:k,4ý,.-.l,
'I!
.1:
1
kiim!!:
li1
;..i
!
ý!
.;
I1.
1
ý
ý1j-1ý1
1
ý-ý-t
;!-'.i
.
..
I
I :II ; !! ,.;. ;i); . , ;1
:
i
ý
;
ji,:
i
i
I
w...
-.
i
---.-.
;1
-Ii;. I!I.0
;:
!;
!,
:I,--.
;.
:i
;
I.
;
, I !..9
Itj 1
I1
,!
.:
:
; .
,
..
l
!
-,.. j
1,
;......
,..
,
:,
I
;
j.
.-;
.i;,!;
i I..,I- .I,.......
wiý,
,
:
izI
1
1
:
I . I. :-.;
!
!
,
.
.
.
.
..
.
:!
4
:
:, :..,..-.---.
-.;lk
I,!, : I : :1ill'
-ýý-j
I
.
:
I
.
.
I
I
!
!
- ý, ... -- - -; :
ý
,
.
I
:
,
,.
:
1
;
1
3
:
;
1*
.....
.
- .,.- ...r..--,
.1
1 t.4.-,-;-..
.. .
- . . . . ý-,I.i
,
iý
. .. i ? iý. -. m, ý-L-, .
t..
I... . .-- ;. I-..- .t.- ý .I.---!.-..-,.,,= ,4ý
...ý.-,-.ýý
: , ..
. P-*-,;7
I+
" .. If-.,.-ý1...
ia. -4 ,4--ý
ý..i..:.. -- .- I-I- .- .4
. i. 1-,..: .- 7.i-f.-. ,ý.A-i,.4I...
..;
...
- IIt'i !- IN
i... -4:
. .. .. ,,-.ý.. , - T, ---ýI..-.-ý ---j-ý-l-,-ý!..-.
. . -....
- -11--t- a
l,- . ; P ,:,.--.,--:,-ý
- ...- ::! -.--ý..-+1,-..-..,..- .. ...
-.---.,-.,.--..
-.....-,: ..........
.!...
.......
ý.,.--f.----. .ý-!--ý--i
'. i;*-!..
. .
ý
I. -1- I - (1:1 - ,-I'..,l...
....:- :I. its-t-,-i Z
--.-- :,c.'..
I.-I
1.
!1.-..
....
:.;...T
. 4-TI.I;-i.--..
J-..ýI. . 4 -r
.f..
. 71-,.j..
4..t.-;
I.j11
.ý. -:.-4
....-.-ýj:..
..-.' - r-Aý
,..-4-:-ý..
-:.....
-. -..ýt..
....
-p
..
:-.-,
.
..
....
ý,
.
47,
..1
i:ýr
.
it
-,.
m
I
!ý
I.;
:
ý,
.
i
.
.
.-:
_!..
.
..--.
,
---,j
.
.
L...
-i
-:
ý_
,-..ý.
T
...
.
ý
I
.:
.
:
..
!
I
m.
............
-.
-iýW
il+ý--ý
----..ý-4-44........
.
! -:..,I-.,ý-i.-;;.l- it -I----Z
.-...ý--,.- 4. - .-.. : ..-...
-.l...
. .. I..
.z------ý(&ý
J-L-Lý
lq-+-,ýI.,
.ý-,:ýý
-.ý.:.,1.I.-.
",-ý-,
.ýA
ý.
-".
,.:".
,
4--ý-4
.,,
-,
1. .... .- .
.
.
,-- .-,
: ;...m.
-;ý-7'. ...;......-.,
..:-,.......
.--t-4...
...-.
. -.
. . Iý
--:
i-.
-4--!.-ý-.
-ý--;.-,-.
--.-.-..
1.
..
...
,..ýý...:
..
-ýý
.....
.......
,
-ý-.,-...--L-.ý
,
j
-,.-.L-;L-.L.44-L
.-..
4
....
j
..
.
-i
I
-1
-,'::
. .----,I -- -L,.-: .- ,.ý -"-., " -.. .. . ý. ýý..;
M- _;I--. [..ý...-.. L-lt:t.t-±
. . ."-' -!7" . . . ý.T- , - -L. --.7.,
:ý.! ... . . f- .:,--.1
,-!
:-. ,--ý- .4#
:
. .. .:- I ,-.
.. ,_4-..;.. .-I
; -.1Ii1
.F
.
i-- .-!-----.Lý
. !,. .. .i - 1; .i 4. . ,
. I- -LI- -_- .: ý-ý--,-.ý.-IP,,,..,.-;;
.:-I.f.
--4,ý.
.f..f,:I..----, . -- t I..= ..;
. .--ý,.i.i-ý-. . .I
. ,-..- ....-...J;-,
.
..
t.-.
,
-:.i
I
.
-,
,
i,....ý
-..
.
_
_
-ýj..-ý
.4
..11
-...--.
ý.ý..
..
4I,
.
l-1
-1-4-4
,
;
.
--ý+
.
;
.
.
I
,,-.-,=
.
:
*,
ýý
-1,
L'.
,
.
.
E'EL.'.
:ý,
- -L.j.
.-I..I.-i= -ý..
1, xlý- .- -- -ý
---ý-.--,Iý--ýý-:-,--.-ý I
1..-ý .-..-:.-.-_
-4"ý.i--L
i Id:7
.-+...
. .- ..,.-:-,
_;_
:i;.
,
I !--.
....
I..
--------..
1..
.....
--:..ý;,
,
-- I...
ý--l. -I -- .7 , ý. I.. a -ý-Ihif+- :, ._j
-!:. .114..;! -I Ii
-.Iý..1..-...
- ..-:..;,
-. ri.i.,ýa. t-ý
I.--4
--.- .=..II : i - . I
----- ý-...ý..
I--.
4.
, -1,IE--:6-.1p:a 1
I ; 7. -,.ýý
..1-ý,
.:.-;L ý,
,..I.--i---trA
-...
; -- 1:--1.4-ý--;+
! .- .--I
-i-ý.....
. iI L-;- ..ý
I, 4-9.
i+-.--;-.-L
...
-;
--ý-4....H.-ýi
.--!-:. ..ý-- - I .. ,.I .ev
.,
-7.
i
.
.
I
I
.
-i.-L-.--;-..
.
-ý--,4
..-....
-..-....---.....-..
,
I
.
;*/.
I
...
.
.
.I
--.
. . I"
!!!
ý
::
,:-,.--r.I-:-+t
.,
ý,
,
ft,
t
..
I--.-... .I..-T---T-iý:,-,t'.. ,.-...!
1.-, ý...,
.
I
..
I
I
L.;
A
.
,,
-ý.,
..
-7,
.
..
I
:.
..
,
-t
. -,.-1-:....
=
ii.:...-"-..-'
I-id,
.-l.-Fý
.11.j.,
,.
,
-1-1-t4-4-,,
t--..ý.-T]
1-i-,..-;-jI--ýý.-..
i .:j-j--r::C7
-!..ý
-.
- .--...-.;.,
-- j. --- -- .4.ý -;ý..I
;
t
,
1.
-,:",.
-..
-4-4-,-.
-,
.
,ý,
.
-.1--L+ý---,
ý.._
I
Is
.+
.
..
..
.
1
-,
+ý--,
..
+
4-ý.
.
-1-t-444-4
,
;
-t---,,-4-L;.
. . . V
I-4
- - .- ! -j-.1-,--., i 1 - 1. .
7ffFTiJH--;.._..+-.-.-.ý
;-. .t.-,.l....:-...I
1::
..:-.-...
...
I.
.--..
1
-,
..
.41--tt"
tt4l
....
i,!,.,
.ý
I- . 71
-.
ý,--.
7 .;...I..i.,
I.
-:
o.--.;
..-- ....
ý--.,1. .4. ...I
- ý-:..!.
...
i
I
I
.
I1.
..
4
.........
.....
.
:.
:
;.
..
.
i
i
I Io
I
.
I
I
I
iý
ýý
i
i
I
l
II
lq....
AA.I
-..-,. 1.1
-...
1:.....
-b
:.ý.
ý.
- ,...:...ý
::--ý .v
.. ..,ý.-ý
-,-.--4,.+ I
ý4-4.
;.ý.!ý-4-ý-Ir4-ý.-....:......ý.-.ý
.iI...'"
ýý--.+t-i-t--- ' -I..-,,--:.
i-.: !:..
.. .;,--.-.;--ý
,
I','j1i'ii1:1ý:!iý:ýiý:-!-
i,!:::; !, iisý;.; :. ,--- ii:; ýt:!ýýJ!ý:!ii!i i.ý ; 1
-'--Iý-.,- - ý- -1- 'L.
-ý+
ýý......
-Lý.
- -1
......
4..-.-L-.ý
L--..
-........j.--..I...;.. ,..;-I
-1
--!- , ,:..,-.--.
-.--...
t.L ,-..--i.-ý ; I , : +ý.
:-i ..: I-...
.....
f- i..
1111F
: ;1.
. ý I4-; I ;-7ý-,, ý !ý-,-.
- . .".
I..... . .. : 1 ,-..ý- I ..-.-- .ýýT
...
4 - -t..I.::..:l,--.,.--t--f
,.. -t ...
Z-; ma, -;.,-. .
I,
, , -.
1.
c..
-.-
ý
II
I
II
--.
:1.
...-ý-
.
L .-- !- II
,-!..ý-ý -:--..-I.-,,4-+
f
ij.-ý-,ý--42,
,-4-..
;.-4
P-TrtI.
,. :-ý+-.
. II , .., ,-X
-- - !-r't-ýV mI.! -..; ---
.-.
/.
. H _:...!ý
.4-.[-..
. 44----ý----,--, I. .- ,-Lý-1-4-i., -ý -;-.i---'-Lý
I
. ...I......
: : 4-4--- - .
- ý-i
ý. . I- .-.. ..I.- +'.-..
,----.
-+--- -m. A.- 4---j,: 4 yý------ý .. -..I ..
I -.T--f--!
:-L,
.- , !-.7-1.
!- -i---.t--I ,...... --;--, , -i .- .,- ,..L.---,: _:.,
,. . . . . - . .4r..
. - ; ý - .....-.!
j
I.,
',-a,.-.,
-ý.
I
..W-..,i-q
it:,-t.
.
iý.-,
ý
I.--.---.
-,-.- ý.
-. I.- -
.. ý
..--
, +:7 -
..
.
---
I.."..iw I., --. 1!.ýW -,
1. ý..:..-.,.
--..,....
.: -. ....
;r... .-!-.... Aý
.....
.r.
-I-ý-.... I..- !.ý.
ý-.-.. .....
-. -..
-.M.-.,lo-ý
-- -- r .:
-...... .
.. .
.. -... "IL.-I...
- - 1.,.---..-..,-.
,,,i&
, i . i--:-ý-,-,
-!-----: !
-''- -- - j-+--.:-!
ý!'Nwl
T..-L..-...
--tI.-4
..,-,.--....`
ý....A
-....
-. . - -.. -II I......
.....
t.--...---!.
-.
1
1.
ý 1.
- --....-I. ý.r.-.:r-4.
-i--r-;I.A....
I I . . . .-...-..,--.----..
- . I ..-- Lý
-,ýýý
r--ý-.... ...- W,
4 -..
:.. ". . ;::. t ..I.-:
;f I....
t,
: . - 1 7ý . z-....
. ; -I : 11,
11, 4
4.4!
.- ý L.....-ý..;..ý
.-..-.-..
. . . - ;ý .L.
..,.-..
ý 1,. -.I ..- ,--.o-j--.-;-ý- .
..
L..f-, ...-;....
.!. ....
ý4 .4.I. w-.:!
- ,;I , . . .i-ý
,-;--ýI -.- . . .. . : .-r.!...
.
- ... :-.,.!.-;..4..,
; ....I--.
!- . ; :
. ý-.
.
.
... .4-I
f.....: :
.
.4-1
.....
4. -,-!..ý.j--.:.,.;-ý
I- ...:...
. .I....:
I. :-:, V
! .:.,
..i ...ý....
: 4 .1 .i.
-,-:.
,.., - - 1 ---,11
. . .. .
. . i
,
ý
ý
.
..,;;;v
I
..
j....
-' ' , !-! - .1 .....
, ...
: , . I - ! --.. ý_ý-ý....;_j--.1.1-:..
.1 .ýýiý . . F; : % . .,
-1..-ý.
,..
.1-:1
.!1.
..-. . . t..;-:. ; .. ., : . -ic
: .
[,ý-ý,TýII
.. .1. .--...
i.
-,
.ý. .- . . I,
-:T-.--I--.-: 4-ý---!'
-
"
[ýý:7.
.-]..-.-".:..1.
:.. -J. ;,.ý:,-'...U.;
il"i-ý-.H
.
...A
,:.....
....
:..!-1.:m
:..
,...L.i-ý
...
L
.ý.:.
;4
.
;
.
:ýH-.............:7,:...j.
a I.I.
I.'1.!..
,-1
-F....I.....;
..
.
",
...
:.
,
i.N
..
,1-.
.!]
- .--.-....I.- ........
.
;-.. . -...
!.I
-.-w1I....;
. ;I..;.:l--.:.i
..I
...tml
.,;..-.;...L.-:..ý..I-.......
.
.-
..4 j
.-.:....
.. ...; : I.... -.: .-.., .1 .
..:.-..
.ý.
.
.. :1
. .. .
Ia. .. .. --. - I .:..:.:-,- ,- . . ,.; ......!
i .
;)
.I .. . .. . ý.... ý..- ........: .;. ýl- r.-iI.
-i:-: . : .; : .. ,
.f --i-1 ;I
.
.. :1. I., ý:.:!.ýI
- I. . .
: .: . ; .ý , ! ý! tj
.. ..
.
.
.
.
.
. . .. . . . .:
,. . : .!
.II.. . ..
1;I ý I.------.-.:-.--14
- I- -.. a.., -...
-.. ý p .
.
I
:!ý
ý ii;
.
% . 11
.. .
1
.:
.
_
.,
.1-..
-ý
ý....
-i.- .- --. ::-:7.:....
V.,
:.
.
.. ; .
:: I rr ii :j
.- . ::
i.
J iýij
.
,
11
.. I ,--.---I-.-!
: .I. . , -1
.: ,
l :1
: -..-...-..-,
: . ."...-,It
.,i!
: .I .
.
, ý,. i
i
;;I M
ý.
-
"roar" - original
/ai/
"roar" - synthesized
Fig.
/g/
..
- synthesized
- synthesized
I
-- :
Fig. 4.5
"Where are you?" - synthesized
Synthesized
Original
Original
Synthesized
/da/
Fig. 4.6
I
_._.
I
_
__
~_
_
---
_
~..~
Synthesized
-=II---
-------
Original
/dJr/
Original
Synthesized
Fig. 4.6 (cont)
-·
-
___
Synthesized
Original
Synthesized
Original
Fig. 4.6 (cont)
T,
t
Original
Synthesized
Fig. 4.7
"Are you a good boy ...
Original
Synthesized
Pig. 4.7 (cont)
o.. or a bad boy?"
1
Table 4.
Tamb
fn of pole and zero
__ _
255 cps (63 3 /4 db)
__~_
d.c. output
21.5 0 C
-0.92v
316 ops
(50db)
21.50
-0.54v
369 cps
(56 db)
21.50
-0.32v
2000 cps
(28 db)
21.50
+0.028v
255 cps
30.50
-3.7v
316 cps
30.50
-2.3v
369 cps
30.5
2000 cps
30.0
0
-1.37v
-0.5v
Table 4.2
Input Impedance:
Output Impedance:
10K in parallel with load
5
plus source
Max. Output Current:
15
ma.at ground
Approx. Predicted Offset:
1 my. with 10
source
Table 4.3
Vowel
s ignal/rnoise
+-47 db
+ 53 db
+ 48 db
- 43 db (no computer)
+ 43 db
+ 50 db
+ 45 db
- 40 db (with computer
+ 31 db
+ 41 db
+ 35 db
- 27 db (speaking
no
ratio
not speaking)
excitation)
References
1. Flanagan,
a
Coding
Formant
Fant, C.G.M.,
1953,
which
Henke,
W.L.,
W.
"Computer
Tract" ,,
S.M.
neering,
MIT,
MIT,
Techniques",
1964.
Signals
from
in Communication
1953.
of a
Terminal
Analog
Thesis, Department of Elec-
1963.
"An Improved Dynamic Analog of the
Thesis,
Department of
Vocal
Electrical Engi-
1963.
Whitman, E.C., Tomlinson R.S.,
Vocal
Rate ',
Control
Synthesizer", S.M.
Whitman, E.C.,
Group,
IVA24,
Research'ý,
of Speech
Synthesis
Jackson, London,
trical Engineering,
Human
J.
System",
28, 1956, 1099-1106.
Low Information
have a
Speech
6.
Compression
Speech
"Soeech Communication
W., "The
Theory, ed.
5.
"Development and Testing of
331-337.
3. Lawrence,
4.
Am.
Soc.
Acoust.
2.
House, A.S.,
J.L.,
Tract
Using
"A Terminal Analog of the
Analog-Digital
unpublished report, Speech
Research Laboratory
Computation
Communications
of Electronics,
MIT,
June
7.·
Tomlinson, R.S.,
Attenuator",
The
Design of a Digitally
Quarterly
Progress
Laboratory of Electronics,
"Dynamic Analog
Rosen, G.,
Thesis,
Department
of
MIT,
Report
Controlled
74, Research
June 1964.
Speech Synthesizer",
Electrical
Sc.D.
Engineering,
MTT,
1960.
9.
Stevens,
cal
K.N.,
Fant, C.G.M.,
Analog of the Vocal Tract",
25, 1953,
10.
Kasowski, S.,
J.
"An
Electri-
Acoust.
Soc.
Am.
734-742.
Flanagan, J.L., "Note on the Design of Terminal
Speech Synthesizers", J. Acoust.
Soc.
Analog
29, 1957,
Am.
306.
11.
Fant CoG.M°,
Acoustic
Theory
Mouton and Company, The Hague,
12.
Fant, C.G.M.,
of
Speech
Production,
1960.
"The Acoustics of Speech",
Proc.
of
the
1959,
ed.
Third International Congress on
Acoustics
Cremer, Elsevier Publishing Co.,
1961, 188-200.
13.
House, A.S.,
Stevens, K.N.,
"An Acoustical Theory
of
Vowel Production and Some of its Implications", Journal
4, Dec.
4, No.
of Speech and Hearing Research
1961,
303-320.
14.
Holbrook, A.,
Fairbanks,
Their Movements",
Journal of
"Some
0.,
Fujimura,
Consonants in the
Report
Speech and
Synthesis
Experiments
Re-
on Stop
uaterly ?rogress
nitial Position ,
Electronics, MIT,
of
Research Laboratory
61,
Hearing
1, March 1962.
search 5, No.
15.
'Diphthong Formants and
G.,
April 1961, 153-162.
16.
Heinz,
J.M.,
Stevens,
K.N., "On
the
Properties
of
Soc.
Am.
Lehiste, I., Peterson, G.E., 'Transitions, Glides,
and
Voiceless Fricative Consonants", J. Acoust.
33, 1961, 589-596.
17.
Diphthongs", J. Acoust.
18.
Delattre,
tic
P.C.,
Loci
Acoust.
and
Soc.
Soc.
Liberman, A.M.,
Am.
Cooper, F.S.,
Transitional Cues
Am.
33, 1961, 268-277.
"Ac ous -
for Consonants", J.
27, 1955, 764-773.
19.
C.G.M., "Acoustical Analysis and
Fant,
of
Synthesis
Speech with Applications to Swedish", Ericsson Technics
1, 1959.
20.
Thaler, G.J. Brown, R.G.,
and Engineering,
Physics
22.
of Combined
Analog-
Final Report ESL-FR-
Computation Techniques'",
Digital
New York, 1958.
McGraw-Hill,
"Application Studies
Hill, F.B.,
of
Mathematics
R.M.,
Redheffer,
I.S.,
Sokolnikoff,
New York, 1960
McGraw-Hill,
back Control Systems,
21.
Analysis and Design of Feed-
165, Electronic Systems Laboratory, MIT, February 1963.
23.
Fujimura,
Acoust.
24.
O.,
Soc.
"Analysis
Am.
of
34, Dec.
Nasal
1962,
K.,
Analog
Synthesizer",
Hecker, M.H.L.,
Articulatory
Am.
26.
1865-1875.
"Synthesis of Nasal Consonants by
Nakata,
Quarterly
Progress
Research Laboratory of Electronics,
25.
Consonants",
34, Feb.
"Studies of Nasal
Speech
Synthesizer",
Terminal
MIT, April 1958.
Consonants with
J.
Acoust.
1962.
Expensive Typewriter Ila,
49,
Report
Memorandum PDP-22,
Computer, Research Laboratory of Electronics,
PDP-1
MIT.
an
Soc.