CHAPTER 7

Speech Coding

peech coders have assumed considerable

importance in communication systems as their performance, to a large extent,
determines the quality of the recovered speech and the capacity of the system. In
mobile communication systems, bandwidth is a precious commodity, and service
providers are continuously met with the challenge of accommodating more users
within a limited allocated bandwidth. Low bit-rate speech coding offers a way to
meet this challenge. The lower the bit rate at which the coder can deliver toll-
quality speech, the more speech channels can be compressed within a given
bandwidth. For this reason, manufacturers and service providers are continu-
ously in search of speech coders that will provide toll quality speech at lower bit
rates.

7.1 Introduction
In mobile communication systems, the design and subjective test of speech
coders has been extremely difficult. Without low data rate speech coding, digital
modulation schemes offer little in the way of spectral efficiency for voice traffic.
To make speech coding practical, implementations must consume little power
and provide tolerable, if not excellent speech quality.
The goal of all speech coding systems is to transmit speech with the highest
possible quality using the least possible channel capacity. This has to be accom-
plished while maintaining certain required levels of complexity of implementa-
tion and communication delay. In general, there is a positive correlation between
coder bit-rate efficiency and the algorithmic complexity required to achieve it.
The more complex an algorithm is, the more its processing delay and cost of
implementation. A balance needs to be struck between these conflicting factors,

361
 362 Ch. 7• Speech Coding

and it is the aim of all speech processing developments to shift the point at which
this balance is made towards ever lower bit rates [Jay92).
The hierarchy of speech coders is shown in Figure 7.1. The principles used
to design and implement the speech coding techniques in Figure 7.1 are
described throughout this chapter.

LPC

Figure 7.1
Hierarchy of speech coders (courtesy of R.Z. Zaputowycz).

Speech coders differ widely in their approaches to achieving signal com-

pression. Based on the means by which they achieve compression, speech coders
are broadly classified into two categories: Waveform Coders and Vocoders. Wave-
form coders essentially strive to reproduce the time waveform of the speech sig-
nal as closely as possible. They are, in principle, designed to be source
independent and can hence code equally well a variety of signals. They have the
advantage of being robust for a wide range of speech characteristics and for noisy
environments. All these advantages are preserved with minimal complexity, and
in general this class of coders achieves only moderate economy in transmission
bit rate. Examples of waveform coders include pulse code modulation (PCM), dif-
ferential pulse code modulation (DPCM), adaptive differential pulse code modu-
lation (ADPCM), delta modulation (DM), continuously variable slope delta
modulation (CVSDM), and adaptive predictive coding (APC) [Del93]. Vocoders on
the other hand achieve very high economy in transmission bit rate and are in
Characteristics of Speech Signats 363

general more complex. They are based on using a priori knowledge about the sig-
nal to be coded, and for this reason, they are, in general, signal specific.

7.2 Characteristics of Speech Signals

Speech waveforms have a number of useful properties that can be exploited
when designing efficient coders [Fla79]. Some of the properties that are most
often utilized in coder design include the nonuniform probability distribution of
speech amplitude, the nonzero autocorrelation between successive speech sam-
ples, the nonflat nature of the speech spectra, the existence of voiced and
unvoiced segments in speech, and the quasiperiodicity of voiced speech signals.
The most basic property of speech waveforms that is exploited by all speech cod-
ers is that they are bandlimited. A finite bandwidth means that it can be time-
discretized (sampled) at a finite rate and reconstructed completely from its sam-
ples, provided that the sampling frequency is greater than twice the highest fre-
quency component in the low pass signal. While the band limited property of
speech signals makes sampling possible, the aforementioned properties allow
quantization, the other most important process in speech coding, to be performed
with greater efficiency.
Probability Density Function (pd/) —- The nonuniform probability den-
sity function of speech amplitudes is perhaps the next most exploited property of
speech. The pdf of a speech signal is in general characterized by a very high
probability of near-zero amplitudes, a significant probability of very high ampli-
tudes, and a monotonically decreasing function of amplitudes between these
extremes. The exact distribution, however, depends on the input bandwidth and
recording conditions. The two-sided exponential (Laplacian) function given in
equation (7.1) provides a good approximation to the long-term pdf of telephone
quality speech signals IJay84].

p(x) = (7.1)

Note that this pdf shows a distinct peak at zero which is due to the exist-
ence of frequent pauses and low level speech segments. Short-time pdfs of speech
segments are also single-peaked functions and are usually approximated as a
Gaussian distribution.
Nonuniform quantizers, including the vector quantizers. attempt to match
the distribution of quantization levels to that of the pdf of the input speech sig-
nal by allocating more quantization levels in regions of high probability and
fewer levels in regions where the probability is low.
Autocorrelation Function (ACF) — Mother very useful property of
speech signals is that there exists much correlation between adjacent samples of
a segment of speech. This implies that in every sample of speech, there is a large
component that is easily predicted from the value of the previous samples with a
364 Ch. 7• Speech Coding

small random error. All differential and predictive coding schemes are based on
exploiting this property.
The autocorrelation function (ACF) gives a quantitative measure of the
closeness or similarity between samples of a speech signal as a function of their
time separation. This function is mathematically defined as [Jay84]
N—Ikl—I
C(k)
=
x(n)x(n+IkI) (7.2)

where x (k) represents the k th speech sample. The autocorrelation function is

often normalized to the variance of the speech signal and hence is constrained to
have values in the range (-1,1) with C (0) = I . signals have an adjacent
sample correlation, C (I) , as high as 0.85 to 0.9.
Power Spec tral Density Function (PSI)) — The nonflat characteristic of
the power spectral density of speech makes it possible to obtain significant com-
pression by coding speech in the frequency domain. The nonflat nature of the
PSD is basically a frequency domain manifestation of the nonzero autocorrela-
tion property. long-term averaged PSD's of speech show that high fre-
quency components contribute very little to the total speech energy. This
indicates that coding speech separately in thiferent frequency bands can lead to
significant coding gain. However, it should be noted that the high frequency com-
ponents, though insignificant in energy are very important carriers of speech
information and hence need to be adequately represented in the coding system.
A qualitative measure of the theoretical maximum coding gain that can be
obtained by exploiting the nonilat characteristics of the speech spectra is given
by the spectral flatness measure (SFM). The SFM is defined as the ratio of the
arithmetic to geometric mean of the samples of the PSD taken at uniform inter-
vals in frequency. Mathematically,
- N n

SFM= k=l (7.3)

N

where, Sk is the k th frequency sample of the PSD of the speech signal. Typi-
cally, speech signals have a long-term SFM value of 8 and a short-time SFM
value varying widely between 2 and 500.
 Quartization Techniques 365

7.3 Quantization Techniques

7.3.1 Uniform Quantization

Quantization is the process of mapping a continuous range of amplitudes of
a signal into a finite set of discrete amplitudes. Quantizers can be thought of as
devices that remove the irrelevancies in the signal, ahd their operation is irre-
versible. Unlike sampling, quantization introduces distortion. Amplitude quanti-
zation is an important step in any speech coding process, and it determines to a
great extent the overall distortion as well as the bit rate necessary to represent
the speech waveform. A quantizer that uses n bits can have M = 2" discrete
amplitude levels. The distortion introduced by any quantization operation is
directly proportional to the square of the step size, which in turn is inversely pro-
portional to the number of levels for a given amplitude range. One of the most
frequently used measures of distortion is the mean square error distortion which
is defined as:
T
MSE=E[(x—fQ(xfl2J = (7.4)

where x (t) represents the original speech signal, and (t) represents the
quantized speech signal. The distortion introduced by a quantizer is often mod-
eled as additive quantization noise, and the performance of a quantizer is mea-
sured as the output signal-to-quantization noise ratio (SQNR). A pulse code
modulation (PCM) coder is basically a quantizer of sampled speech amplitudes.
PCM coding, using 8 bits per sample at a sampling frequency of 8 kHz, was the
first digital coding standard adopted for commercial telephony. The SQNR of a
PCM encoder is related to the number of bits used for encoding through the fol-
lowing relation:
(SQNR)dB = 6.02n + a (7.5)
where a = 4.77 for peak SQNR and a = 0 for the average SQNR. The above equa-
tion indicates that with every additional bit used for encoding, the output SQNR
improves by 6 dB.

7.3.2 Nonuniform Guantization

The performance of a quantizer can be improved by distributing the quanti-
zation levels in a more efficient manner. Nonuniform quantizers distribute the
quantization levels in accordance with the pdf of the input waveform. For an
input signal with a pdfp(x),the mean square distortion is given by

D = E[(x—fQ(xfl2l [x_fQ(x)12p(x)dx
= f (7.6)
366 Ch. 7 Speech Cod:ng

where fQ (x) is the output of the quantizer. From the above equation, it is clear
that the total distortion can be reduced by decreasing the quantization noise,
[x (x) 12, where the pdf, p (x). is large. This means that quantization levels
need to be concentrated in amplitude regions of high probability.
To design an optimal nonuniform quantizer, we need to determine the
quantization levels which will minimize the distortion of a signal with a given
pdf. The Lloyd-Max algorithm provides a method to determine the opti-
mum quantization levels by iteratively changing the quantization levels in a
manner that minimizes the mean square distortion.
A simple and robust implementation of a nonuniform quantizer used in
commercial telephony is the logarithmic quantizer. This quantizer uses fine
quantization steps for the frequently occurring low amplitudes in speech and
uses much coarser steps for the less frequent, large amplitude excursions. Differ-
ent companding techniques known as p-law and A-law companding are used in
the U.S and Europe, respectively.
Nonuniform quantization is obtained by first passing the analog speech sig-
nal through a compression (logarithmic) amplifier, and then passing the com-
pressed speech into a standard uniform quantizer. In U.S. -law companding,
weak speech signals are amplified where strong speech signals are compressed.
Let the speech voltage level into the compander be w (t) and the speech output
voltage be t'0 (t) Following [Smi57],
-

(tfl = ln( I

(7.7)
In ( I + p3
where p is a positive constant and has a value typically between 50 and 300. The
peak value of w (1) is normalized to 1.
In Europe, A-law companding is used [Cat691, and is defined by
( Aiw(tfl
l+lnA A
v U) = (7.8)
I ± in (tfl) i< k
l+]nA
C

Example 7.1
Let the input signal to a quantizer have a probability density function (pdf) as
shown in Figure E7.1. Assume the quantization levels to be (1. 3, 5, 71. Com-
pute the mean square error distortion at the quantizer output and the output
signal-to-distortion ratio. How would you change the distribution of quantiza-
tion levels to decrease the distortion? For what input pdf would this quantizer
be optimal?
Quantization Techniques 367

Solution to Example 7.1

p(x

1/4

Figure E7. 1 Probability density function of the input signal

From Figure E7.1, the pdf of the input signal can be recognized as:
p(x) =

p(x) = 0 elsewhere
Given the quantization levels to be (1, 3, 5. 7} we can define the quantization
boundaries as (0, 2, 4, 6, 8}.

Mean square error distortion = D

= f (x
-
(x) } p (x) dx

D J(x--])2p(x)dx+f(x_3)2p(x)dx+J(x_5)p(x)dx

±J(x 7Yp(x)dx

This expression evaluates to 0.333.

S

Signal Power = E[x2j = Jp(x)x2dx .x )dx = 32

Signal to distortion ratio = Olog [E [xi / D1 = 10kg (32 0.333)

I

= 19.82dB
To minimize the distortion we need to concentrate the quantization levels in
regions of higher probability Since the input signal has a greater probability of
higher amplitude levels than lower amplitudes, we need to place the quantiza-
tion levels closer (i.e., more quantization levels) at amplitudes close to 8 and
farther (i.e., less quantization levels) at amplitudes close to zero.
Since this quantizer has quantization levels uniformly distributed, this would
be optimal for an input signal with a uniform pdf.
 368 Ch. 7• Speech Coding

7.3.3 Adaptive Quantization

noted earlier, there is a distinction between the long term and short
As
term pdf of speech waveforms. This fact is a result of the nonstationarity of
speech signals. The time-varying or nonstationary nature of speech signals
results in a dynamic range of 40 dB or more. An efficient way to accommodate
such a huge dynamic range is to adopt a time varying quantization technique.
An adaptive quantizer varies its step size in accordance to the input speech sig-
nal power. Its characteristics shrink and expand in time like an accordion. The
idea is illustrated in Figure 7.2 through two snapshots of the quantizer charac-
teristics at two different tines. The input power level of the speech signal varies
slowly enough that simple adaptation algorithms can be easily designed and
implemented. One simple adaptation strategy would be to make the step size,
Ak, of the quantizer at any given sampling instant, proportional to the quantizer
output fQ at the preceding sampling instant as shown below.
Since the adaptation follows the quantizer output rather than the input,
step size information need not be explicitly transmitted but can be recreated at
the receiver.

fQ(x) fç(x)

(a) (b)

Figure 7.2
Adaptive quantizer characteristics (a) when the input signaL has a low amplitude swing and (b) when
the input signal has a large amplitude swing.

7.3.4 Vector Quantization

Shannon's Rate-Distortion Theorem [Sha48] states that there exists a map-
ping from a source waveform to output code words such that for a given distor-
tion D, .11(D) bits per sample are sufficient to reconstruct the waveform with an
average distortion arbitrarily close to D. Therefore, the actual rate R has to be
greater than 11(D). The function 11(D), called the rate-distortion function, repre-
sents a fundamental limit on the achievable rate for a given distortion. Scalar
 Adaptive Differential Pulse Code Modulation 369

quantizers do not achieve performance close to this information theoretical limit.

Shannon predicted that better performance can be achieved by coding many
samples at a time instead of one sample at a time.
Vector quantization (VQ) [Gra84] is a delayed-decision coding technique
which maps a group of input samples (typically a speech frame), called a vector,
to a code book index. A code book is set up consisting of a finite set of vectors cov-
ering the entire anticipated range of values. In each quantizing interval, the
code-book is searched and the index of the entry that gives the best match to the
input signal frame is selected. Vector quantizers can yield better performance
even when the samples are independent of one another. Performance is greatly
enhanced if there is strong correlation between samples in the group.
The number of samples in a block (vector) is called the dimension L of the
vectorquantizer. The rate R of the vector quantizer is defined as:
log2n
R
= L
bus/sample (7.9)
where n is the size of the VQ code book. R may take fractional values also. All
the quantization principles used in scalar quantization apply to vector quantiza-
tion as a straightforward extension. Instead of quantization levels, we have
quantization vectors, and distortion is measured as the squared Euclidean dis-
tance between the quantization vector and the input vector.
Vector quantization is known to be most efficient at very low bit rates
(R = 0.5 bits/sample or less). This is because when B is small, one can afford to
use a large vector dimension L, yet have a reasonable size, 2RL, of the VQ
code book. Use of larger dimensions tends to bring out the inherent capability of
VQ to exploit the redundancies in the components of the vector being quantized.
Vector quantization is a computationally intensive operation and, for this reason,
is not often used to code speech signals directly. However, it is used in many
speech coding systems to quantize the speech analysis parameters like the linear
prediction coefficients, spectral coefficients, filter bank energies, etc. These sys-
tems use improved versions of VQ algorithms which are more comput.ationally
efficient such as the multistage VQ, tree-structured VQ, and shape-gain VQ.

7.4 Adaptive Diflerential Pulse Code Modulation

Pulse code modulation systems do not attempt to remove the redundancies
in the speech signal. Adaptive differential pulse code modulation (ADPCM)
[Jay84] is a more efficient coding scheme which exploits the redundancies
present in the speech signal. As mentioned earlier, adjacent samples of a speech
waveform are often highly correlated. This means that the variance of the differ-
ence between adjacent speech amplitudes is much smaller than the variance of
the speech signal itself ADPCM allows speech to be encoded at a bit rate of 32
kbps, which is half the standard 64 kbps PCM rate, while retaining the same
 370 Ch. 7. Speech Coding

voice quality. Efficient algorithms for ADPCM have been developed and stan-
dardized. The CCIII' standard G.721 ADPCM algorithm for 32 kbps speech cod-
ing is used in cordless telephone systems like CT2 and DEC11
In a differential PCM scheme, the encoder quantizes a succession of adja-
cent sample differences, and the decoder recovers an approximation to the origi-
nal speech signal by essentially integrating quantized adjacent sample
differences. Since the quantization error variance for a given number of bits/
sample R, is directly proportional to the input variance, the reduction obtained
in the quantizer input variance leads directly to a reduction of reconstruction
error variance for a given value of R.
In practice, ADPCM encoders are implemented using signal prediction
techniques. Instead of encoding the difference between adjacent samples, a lin-
ear predictor is used to predict the current sample. The difference between the
predicted and actual sample called the prediction error is then encoded for trans-
mission. Prediction is based on the knowledge of the auuworrelation properties
of speech.
Figure 7.3 shows a simplified block diagram of an ADPCM encoder used in
the CT2 cordless telephone system [Det89J. This encoder consists of a quantizer
that maps the input signal sample onto a 4-bit output sample. The ADPCM
encoder makes best use of the available dynamic range of 4 bits by varying its
step size in an adaptive manner. The step size of the quantizer depends on the
dynamic range of the input which is speaker dependent and varies with time.
The adaptation is, in practice, achieved by normalizing the input signals via a
scaling factor derived from a prediction of the dynamic range of the current
input. This prediction is obtained from two components: a fast component for sig-
nals with rapid amplitude fluctuations and a slow component for signals that
vary more slowly. The two components are weighted to give a single quantization
scaling factor. It should be noted that the two feedback signals that drive the
algorithm — (k), the estimate of the input signal and y(k), the quantization
scaling factor — are ultimately derived solely from 1(k), the transmitted 4-bit
ADPCM signal. The ADPCM encoder at the transmitter and the ADPCM
decoder, at the receiver, are thus driven by the same control signals, with decod-
ing simply the reverse of encoding.

Example 7.2
In an adaptive PCM system for speech coding, the input speech signal is sam-
pled at 8 kHz, and each sample is represented by 8 bits. The quantizer step
size is recomputed every 10 ms, and it is encoded for transmission using 5 bits.
Compute the transmission bit rate of such a speech coder. What would be the
average and peak SQNR of this system?
 Frequency Domain Coding of Speech 371

&k)

Figure 7.3
Block diagram of ADPCM encoder.

Solution to Example 7.2

Given:
Sampling frequency = = 8 kHz
Number of bits per sample = n = 8 bits
Number of information bits per second = 8000 x 8 = 64000 bits
Since the quantization step size is recomputed every 10 ms, we have 100 step
size 8amples to be transmitted every second.
Therefore, the number of overhead bits = 100 x 5 = 500 bitsls
Therefore, the effective transmission bit rate = 64000 + 500 = 64.5 kbps.
The noise ratio depends only on the number of bits used
to quantize the samples.
Peak signal to quantization noise ratio in dB
= 6.0Th +477 = (6.02 xY = $2.93 dB.
Average signal to noise ratio in dB = 6.02n = 48.16 dB.

7.5 Frequency Domain Codflng of Speech

Frequency domain coders ETri79] are a class of speech coders which take
advantage of speech perception and generation models without making the algo-
rithm totally dependent on the models used. In this class of coders, the speech
signal is divided into a set of frequency components which are quantized and
encoded separately. In this way different frequency bands can be preferentially
372 Ch. 7• Speech Coding

encoded according to some perceptual criteria for each band, and hence the
quantization noise can be contained within bands and prevented from creating
harmonic distortions outside the band. These schemes have the advantage that
the number of bits used to encode each frequency component can be dynamically
varied and shared among the different bands.
Many frequency domain coding algorithms, ranging from simple to complex
are available. The most common types of frequency domain coding include sub-
band coding (SBC) and block transform coding. While a sub-band coder divides
the speech signal into many smaller sub-bands and encodes each sub-band sepa-
rately according to some perceptual criterion, a transform coder codes the short-
time transform of a windowed sequence of samples and encodes them with num-
ber of bits proportional to its perceptual significance.

7.5.1 Sub-band Coding

Sub-band coding can be thought of as a method of controlling and thstribut-
ing quantization noise across the signal spectrum. Quantization is a nonlinear
operation which produces distortion products that are typically broad in spec-
trum. The human ear does not detect the quantization distortion at all frequen-
cies equally well. It is therefore possible to achieve substantial improvement in
quality by coding the signal in narrower bands. In a sub-band coder, speech is
typically divided into four or eight sub-bands by a bank of filters, and each sub-
band is sampled at a bandpass Nyquist rate (which is lower than the original
sampling rate) and encoded with different accuracy in accordance to a perceptual
criteria. Band-splitting can be done in many ways. One approach could be to
divide the entire speech band into unequal sub-bands that contribute equally to
the articulation index. One partitioning of the speech band according to this
method as suggested by Crochiere, et al, [Cro76] is given below.
Sub-band Number Frequency Range p

1 200-700 Hz
2 700-1310 Hz
3 13 10-2020 Hz
4 2020-3200 Hz

Another way to split the speech band would be to divide it into equal with'
sub-bands and assign to each sub-band number of bits proportional to perceptu.
significance while encoding them. Instead of partitioning into equal width band
octave band splitting is often employed. As the human ear has an exponential]
decreasing sensitivity to frequency, this kind of splitting is more in tune with V
perception process.
There are various methods for processing the sub-band signals. One obv
ous way is to make a low pass translation of the sub-band signal to zero Fr
quency by a modulation process equivalent to single sideband modulation.
Frequency Domain Coding of Speech 373

kind of translation facilitates sampling rate reduction and possesses other bene-
fits that accrue from coding low-pass signals. Figure 7.4 shows a simple means of
achieving this low pass translation. The input signal is filtered with a bandpass
filter of width for the ii th band. is the lower edge of the band and is
the upper edge of the band. The resulting signal (t) is modulated by a cosine
wave cos and filtered using a low pass filter (t) with bandwidth
(0— The resulting signal rn (() corresponds to the low pass translated ver-
sion of (t) and can be expressed as
= øhn(t) (7.10)
where 0 denotes a convolution operation. The signal (t) is sampled at a rate
This signal is then digitally encoded and multiplexed with encoded signals
from other channels as shown in Figure 7.4. At the receiver the data is demulti-
plexed into separate channels, decoded, and bandpass translated to give the esti-
mate of (t) for the ii th channel.
The low pass translation technique is straightforward and takes advantage
of a bank of nonoverlapping bandpass filters. Unfortunately, unless we use
sophisticated bandpass filters, this approach will lead to perceptible aliasing
effects. Estaban and Galand proposed [Est77j a scheme which avoids this incon-
venience even with quasiperfect, sub-band splitting. Filter banks known as
quadrature mirror filters (QMF) are used to achieve this. By designing a set of
mirror filters which satisfy certain symmetry conditions, it is possible to obtain
perfect alias cancellation. This facilitates the implementation of sub-band coding
without the use of very high order filters. This is particularly attractive for real
time implementation as a reduced filter order means a reduced computational
load and also a reduced latency.
Sub-band coding can be used for coding speech at bit rates in the range 9.6
to 32 kbps. In this range, speech quality is roughly equivalent to that of
OPCM at an equivalent bit rate. In addition, its compleiity and relative speech
•uality at low bit rates make it particularly advantageous for coding below about
6 kbps. However, the increased complexity of sub-band coding when compared
other higher bit rate techniques does not warrant its use at bit rates greater
sian about 20 kbps. The CD-900 cellular telephone system uses sub-band coding
'r speech compression.

Example 7.3
Consider a sub-band coding scheme where the speech bandwidth is partitioned
into 4 bands. Table below gives the corner frequencies of each band along with
the number of bits used to encode each band. Assuming that no side informa-
tion need be transmitted, compute the minimum encoding rate of this SBC
coder.
374 Ch. 7 . Speech Coding

Encoder

sj't)

Decoder

s (t)

Figure 7.4
Block diagram of a sub-band coder and decoder.

Sub-band Number Band (Hz) # of Encoding Bits

1 225-450 4
2 450-900 3

1000-1500 2
1800 - 2700
1

4 1
-j
Solution to Example 7.3
Given:
Number of sub-bands = N =4
 Frequency Domain Coding ol Speech 375

For perfect reconstruction of the band-pass signals, they need to be sampled at

a Nyquist rate equal to twice the bandwidth of the signal. Therefore, the differ-
ent sub-bands need to be sampled at the following rates:
Sub-band 1 = 2 x (450—225) = 450 samples/s
Sub-band 2 = 2 x (900—450) = 900 samples/s
Sub-band 3 = 2 x (1500—1000) = 1000 samples/s
Sub-band 4 = 2 x (2700— 1800) = 1800 samples/s
Now, total encoding rate is
= 8300 bits/s=8.3kbps

7.5.2 AdaptIve Transform Coding

Adaptive Transform Coding (ATC) [0we93] is another frequency domain
technique that has been successfully used to encode speech at bit rates in the
range 9.6 kbps to 20 kbps. This is a more complex technique which involves block
transformations of windowed input segments of the speech waveform. Each seg-
ment is represented by a set of transform coefficients, which are separately
quantized and transmitted. At the receiver, the quantized coefficients are
inverse transformed to produce a replica of the original input segment.
One of the most attractive and frequently used transforms for speech cod-
ing is the discrete cosine transform (DCT). The DCT of a N-point sequence x (n)
is defined as

= x(n)g (k) k = 0, 1,2 N—I (7.11)

where g(0) = I and g(k) = [2, k = 1,2 ,N—1. The inverse DCT is
defined as:

x (n) = (k)g (k) cos [ (2n±i)kn] n = 0, 1,2 N- 1 (7.12)

In practical situations the DCT and IDCT are not evaluated directly using
the above equations. Fast algorithms developed for computing the DCT in a corn-
putationally efficient manner are used.
Most of the practical transform coding schemes vary the bit allocation
among different coefficients adaptively from frame to frame while keeping the
total number of bits constant. This dynamic bit allocation is controlled by tine-
varying statistics which have to be transmitted as side information. This consti-
tutes an overhead of about 2 kbps. The frame of N samples to be transformed or
inverse-transformed is accumulated in the buffer in the transmitter and receiver
respectively. The side information is also used to determine the step size of the
various coefficient quantizers. In a practical system, the side information trans-
mitted is a coarse representation of the log-energy spectrum. This typically con-
sists of L frequency points, where L is in the range 15-20, which are computed
376 Ch. 7• Speech Coding

by averaging sets of IV/L adjacent squared values of the transform coefficients

X (k) . At the receiver, an N -point spectrum is reconstructed from the L -point
spectrum by geometric interpolation in the log-domain. The number of bits
assigned to each transform coefficient is proportional to its corresponding spec-
tral energy value.

7.6 Vocoders
Vocoders are a class of speech coding systems that analyze the voice signal
at the transmitter, transmit parameters derived from the analysis, and then syn-
thesize the voice at the receiver using those parameters. All vocoder systems
attempt to model the speech generation process as a dynamic system and try to
quantify certain physical constraints of the system. These physical constraints
are used to provide a parsimonious description of the speech signal. Vocoders are
in general much more complex than the waveform coders and achieve very high
economy in transmission bit rate. However, they are less robust, and their per-
formance tends to be talker dependent. The most popular among the vocoding
systems is the linear predictive coder (LPC). The other vocoding schemes include
the channel vocoder, formant vocoder, cepstrum vocoder and voice excited
vocoder.
Figure 7.5 shows the traditional speech generation model that is the basis
of all vocoding systems [F1a791. The sound generating mechanism forms the
source and is linearly separated from the intelligence modulating vocal tract fil-
ter which forms the system. The speech signal is assumed to be of two types:
voiced and unvoiced. Voiced sound (5n", pronunciations) are a result of
quasiperiodic vibrations of the vocal chord and unvoiced sounds (7', "s", "sh" pro-
nunciations) are fricatives produced by turbulent air flow through a constriction.
The parameters associated with this model are the voice pitch, the pole frequen-
cies of the modulating filter, and the corresponding amplitude parameters. The
pitch frequency for most speakers is below 300 Hz, and extracting this informa-
tion from the signal is very difficult. The pole frequencies correspond to the reso-
nant frequencies of the vocal tract and are often called the fonnants of the
speech signal. For adult speakers, the formants are centered around 500 Hz,
1500 Hz, 2500 Hz, and 3500 Hz. By meticulously adjusting the parameters of the
speech generation model, good quality speech can be synthesized.

7.6.1 Channel Vocoders

The channel vocoder was the first among the analysis-synthesis systems of
speech demonstrated practically. Channel vocoders are frequency domain vocod-
ers that determine the envelope of the speech signal for a number of frequency
bands and then sample, encode, and multiplex these samples with the encoded
outputs of the other filters. The sampling is done synchronously every 10 ms to
 ______ _____

Vocoders 377

r —
Second Source

speech
/ output

Pulse
Source

L J

speech r'J
input

Figure 7.5 Transmission

Speech generation model.

30 ms. Along with the energy information about each band, the voiced/unvoiced
decision, and the pitch frequency for voiced speech are also transmitted.

7.6.2 Formant Vocoders

The formant vocoder [Bay73} is similar in concept to the channel vocoder.
Theoretically, the formant vocoder can operate at lower bit rates than the chan-
nel vocoder because it uses fewer control signals. Instead of sending samples of
the power spectrum envelope, the formant vocoder attempts to transmit the
tions of the peaks (formants) of the spectral envelope. Typically, a formant
vocoder must be able to identify at least three formants for representing the
speech sounds, and it must also control the intensities of the formants.
Formant vocoders can reproduce speech at bit rates lower than 1200 bits/s.
However, due to difficulties in accurately computing the location of formants and
forwant transitions from human speech, they have not been very successful.

7.6.3 Cepstrum Vocoders

The Cepstrum vocoder separates the excitation and vocal tract spectrum by
inverse Fourier transforming of the log magnitude spectrum to produce the cep-
strum of the signal. The low frequency coefficients in the cepstrum correspond to
 378 CIt 7• Speech Coding

the vocal tract spectral envelope, with the high frequency excitation coefficients
forming a periodic pulse train at multiples of the sampling period. Linear filter-
ing is perfonned to separate the vocal tract cepstral coefficients from the excita-
tion coefficients. In the receiver, the vocal tract cepstral coefficients are Fourier
transformed to produce the vocal tract impulse response. By convolving this
impulse response with a synthetic excitation signal (random noise or periodic
pulse train), the original speech is reconstructed.

7.6.4 Voice-Excited Vocoder

Voice-excited vocoders eliminate the need for pitch extraction and voicing
detection operations. This system uses a hybrid combination of PCM transinis-
sion for the low frequency band of speech, combined with channel vocoding of
higher frequency bands. A pitch signal is generated at the synthesizer by rectilS'-
ing, bandpass filtering, and clipping the baseband signal, thus creating a spec-
trally flat signal with energy at pitch harmonics. Voice excited vocoders have
been designed for operation at 7200 bits/s to 9600 bits/s, and their quality is typ-
ically superior to that obtained by the traditional pitch excited vocoders.

7.7 Linear Predictive Coders

7.7.1 LPC Vocoders

Linear predictive coders (LPC) [Sch85a] belong to the time domain class of
vocoders. This class of vocoders attempts to extract the significant features of
speech from the time waveform. Though LPC coders are computationally inten-
sive, they are by far the most popular among the class of low bit rate vocoders.
With LPC it is possible to transmit good quality voice at 4.8 kbps and poorer
quality voice at even lower rates.
The linear predictive coding system models the vocal tract as an all pole lin-
ear filter with a transfer function described by

H(z) = (7.13)
I +

where C is a gain of the filter and £' represents a unit delay operation. The
excitation to this filter is either a pulse at the pitch frequency or random white
noise depending on whether the speech segment is voiced or unvoiced. The coeffi-
cients of the all pole filter are obtained in the time domain using linear predic-
tion techniques [Mak75]. The prediction principles used are similar to those in
ADPCM coders. However, instead of transmitting quantized values of the error
signal representing the difference between the predicted and actual wavefonn,
the LPC system transmits only selected characteristics of the error signal. The
 Linear Predictive Coders 379

parameters include the gain factor, pitch information, and the voiced/unvoiced
decision information, which allow approximation of the correct error signal. At
the receiver, the received information about the error signal is used to determine
the appropriate excitation for the synthesis filter. That is, the error signal is the
excitation to the decoder. The synthesis filter is designed at the receiver using
the received predictor coefficients. In practice, many LPC coders transmit the fil-
ter coefficients which already represent the error signal and can be directly syn-
thesized by the receiver. Figure 7.6 shows a block diagram of an LPC system
EJay86].
Determination of Predictor coefficients — The linear predictive coder
uses a weighted sum of p past samples to estimate the present sample, where p
is typically in the range of 10-15. Using this technique, the current sample s,,
can be written as a linear sum of the immediately preceding samples s,, k

Sn (7.14)

where, en is the prediction error (residual). The predictor coefficients are calcu-
lated to minimize the average energy E in the error signal that represents the
difference between the predicted and actual speech amplitude.
N N p 2

E= = (715)

where a0 = -1. T5rpically the error is computed for a time window of 10 ms

which corresponds to a value of N = 80. To minimize E with respect to am , it is
required to set the partial derivatives equal to zero.
N p
=0 (7.16)
=

p N
= =0 (7.17)
k = On
The inner summation can be recognized as the correlation coefficient Crm
and hence the above equation can be rewritten as

=0 (7.18)
k=0
After determining the correlation coefficients equation (7.18) can be
used to determine the predictor coefficients. Equation (7.18) is often expressed in
matrix notation and the predictor coefficients calculated using matrix inversion.
A number of algorithms have been developed to speed up the calculation of pre-
0

Figure 7.6
Block diagram of a LPC coding system.
 _______

Linear Predictive Coders 381

dictor coefficients. Normally, the predictor coefficients are not coded directly; as
they would require 8 bits to 10 bits per coefficient for accurate representation
[De193]. The accuracy requirements are lessened by transmitting the reflection
coefficients (a closely related parameter) which have a smaller dynamic range.
These reflection coefficients can be adequately represented by 6 bits per coeffi-
cient. for a 10th order predictor, the total number of bits assigned to the
model parameters per frame is 72, which includes 5 bits for a gain parameter
and 6 bits for the pitch period. If the parameters are estimated every 15 ms to 30
ms, the resulting bit rate is in the range of 2400 bps to 4800 bps. The coding of
the reflection coefficient can be further improved by performing a nonlinear
transformation of the coefficients prior to coding. This nonlinear transformation
reduces the sensitivity of the reflection coefficients to quantization errors. This is
normally done through a log-area ratio (LAB) transform which performs an
inverse hyperbolic tangent mapping of the reflection coefficients, (k)
rl÷R (k)i
LAR 'k) —— tanh-''R
k 'k')lo
I i
I (719)
Various LPC schemes differ in the way they recreate the error signal (exci-
tation) at the receiver. Three alternatives are shown in Figure 7.7 [Luc891. The
first one shows the most popular means. It uses two sources at the receiver, one
of white noise and the other with a series of pulses at the current pitch rate. The
selection of either of these excitation methods is based on the voiced/unvoiced
decision made at the transmitter and communicated to the receiver along with
the other information. This technique requires that the transmitter extract pitch
frequency information which is often very difficult. Moreover, the phase coher-
ence between the harmonic components of the excitation pulse tends to produce
a buzzy twang in the synthesized speech. These problems are mitigated in the
other two approaches: Multi-pulse excited LPC and stochastic or code excited
LPC.

7.7.2 MultI-pulse Excited LPC

Atal showed [Ata86J that, no matter how well the pulse is positioned, exci-
tation by a single pulse per pitch period produces audible distortion. He there-
fore suggested using more than one pulse, typically eight per period, and
adjusting the individual pulse positions and amplitudes sequentially to mini-
mize a spectrally weighted mean square error. This technique called the multi-
pulse excited LPC (MPE-LPC) results in better speech quality, not only because
the prediction residual is better approximated by several pulses per pitch period,
but also because the multi-pulse algorithm does not require pitch detection. The
number of pulses used can be reduced, in particular for high pitched voices, by
incorporating a linear filter with a pitch loop in the synthesizer.
 382 Ch. 7. Speech Coding

Vocoder

Filter
I

Multipulse Excitation

Synthesis
Filter
Stochastic Exciralion

Figure 7.7
LPC excitation methods.

7.7.3 Code-Excited LPC

In this method, the coder and decoder have a predetermined code book of
stochastic (zero-mean white Gaussian) excitation signals [Sch85bI. For each
speech signal the transmitter searches through its code book of stochastic sig-
nals for the one that gives the best perceptual match to the sound when used as
an excitation to the LPC filter. The index of the code book where the best match
was fomid is then transmitted. The receiver uses this index to pick the correct
excitation signal for its synthesizer filter. The code excited LPC (CELP) coders
are extremely complex and can require more than 500 million multiply and add
operations per second. They can provide high quality even when the excitation is
coded at only 0.25 bits per sample. These coders can achieve transmission bit
rates as low as 4.8 kbps.
Figure 7,8 illustrates the procedure for selecting the optimum excitation
signal. The procedure is best illustrated through an example. Consider the cod-
ing of a short 5 ms block of speech signal. At a sampling frequency of 8 kHz, each
block consists of 40 speech samples. A bit rate of 114 bit per sample corresponds
to 10 bits per block. Therefore, there are 2b0 = 1024 possible sequences of length
40 for each block. Each member of the code book provides 40 samples of the exci-
tation signal with a scaling factor that is changed every 5 ms block. The scaled
samples are passed sequentially through two recursive filters, which introduce
 _________ _____ ______________

Linear Predictive Coders 383

voice periodicity and adjust the spectral envelope. The regenerated speech sam-
ples at the output of the second filter are compared with samples of the original
speech signal to form a difference signal. The difference signal represents the
objective error in the regenerated speech signal. This is further processed
through a linear filter which amplifies the perceptually more important frequen-
cies and attenuates the perceptually less important frequencies.
original speech

excitation +

Predictor F

Perceptual
Averaging error
perceptual 4 . 4 .
I

Weighting
. .

error circuit circuit

Filter

Figure 7.8
Block diagram illustrating the CELP code book search.

Though computationally intensive, advances in DSP and VLSI technology

have made real-time implementation of CELP codecs possible. The CDMA digi-
tal cellular standard (15-95) proposed by QUALCOMM uses a variable rate
CELP codec at 1.2 to 14.4 kbps. In 1995, QUALCOMM proposed QCELP13, a
13.4 kbps CELP coder that operates on a 14.4 kbps channel.

7.7.4 Residual Excited LPC

The rationale behind the residual excited LPC (RELP) is related to that of
the DPCM technique in waveform coding [De193]. In this class of LPC coders,
after estimating the model parameters (LP coefficients or related parameters)
and excitation parameters (voiced/unvoiced decision, pitch, gain) from a speech
frame, the speech is synthesized at the transmitter and subtracted from the ong-
ma! speech signal to from a residual signal. The residual signal is quantized.
coded, and transmitted to the receiver along with the LJPC model parameters. At
the receiver the residual error signal is added to the signal generated using the
model parameters to synthesize an approximation of the original speech signal.
The quality of the synthesized speech is improved due to the addition of the
residual error. Figure 7.9 shows a block diagram of a simple RELP codec.
 384 Oh. 7 . Speech Coding

+
s( n)

Encoded
output

Figure 7.9
Block diagram of a kELP encoder.

7.8 Choosing Speech Codecs for Mobile Communications

Choosing the right speech codec is an important step in the design of a digi-
tal mobile communication system [Gow93]. Because of the limited bandwidth
that is available, it is required to compresS speech to maximize the number of
users on the system. A balance must be struck between the perceived quality of
the speech resulting from this compression and the overall system cost and
capacity. Other criterion that must be considered include the end-to-end encod-
ing delay, the algorithmic complexity of the coder, the d.c. power requirements,
compatibility with existing standards, and the robustness of the encoded speech
to transmission errors.
As seen in Chapter 3 and 4, the mobile radio channel is a hostile transinis-
sion medium beset with problems such as fading, multipath, and interference. It
is therefore important that the speech codec be robust to transmission errors.
Depending on the technique used, different speech coders show varying degree of
immunity to transmission errors. For example, under same bit error rate condi-
tions, 40 kbps adaptive delta modulation (ADM) sounds much better than 56
kbps log-PCM [Ste93]. This does not mean that decreasing bit rate enhances the
coder robustness to transmission errors. On the contrary; as speech signals are
represented by fewer and fewer bits, the information content per bit increases
and hence need to be more safely preserved. Low bit rate vocoder type codecs,
which do parametric modeling of the vocal tract and auditory mechanisms, have
some bits carrying critical information which if corrupted would Lead to unac-
ceptable distortion. While transmitting low bit rate encoded speech, it is impera-
Choosing Speech Codecs for Mobile Comn,unications 385

tive to determine the perceptual importance of each bit and group them
according to their sensitivity to errors. Depending on their perceptual signifi-
cance, bits in each group are provided with different levels of error protection
through the use of different forward error correction (FEC) codes.
The choice of the speech coder will also depend on the cell size used. When
the cell size is sufficiently small such that high spectral efficiency is achieved
through frequency reuse, it may be sufficient to use a simple high rate speech
codec. In the cordless telephone systems like CT2 and DECT. which use very
small cells (microcells), 32 kbps ADPCM coders are used to achieve acceptable
performance even without channel coding and equalization. Cellular systems
operating with much larger cells and poorer channel conditions need to use error
correction coding, thereby requiring the speech codecs to operate at lower bit
rates. In mobile satellite communications, the cell sizes are very large and the
available bandwidth is very small. In order to accommodate realistic number of
users the speech rate must be of the order of 3 kbps, requiring the use of vocoder
techniques [Ste93].
The type of multiple access technique used, being an important factor in
determining the spectral efficiency of the system, strongly influences the choice
of speech codec. The US digital TDMA cellular system (IS-54) increased the
capacity of the existing analog system (AMPS) threefold by using an 8 kbps
VSELP speech codec. CDMA systems, due to their innate interference rejection
capabilities and broader bandwidth availability, allow the use of a low bit rate
speech codec without regard to its robustness to transmission errors. Transmis-
sion errors can be corrected with powerful FEC codes, the use of which, in CDMA
systems, does not effect bandwidth efficiency very significantly.
The type of modulation employed also has considerable impact on the
choice of speech codec. For example, using bandwidth-efficient modulation
schemes can lower the bit rate reduction requirements on the speech codec, and
vice versa. Table 7.1 shows a listing of the types of speech codecs used in various
digital mobile communication systems.

Example 7.4
A digital mobile communication system has a forward channel frequency band
ranging between 810 MHz to 826 MHz and a reverse channel band between
940 MHz to 956 MHz. Assume that 90 per cent of the band width is used by
traffic channels. It is required to support at least 1150 simultaneous calls using
FDMA The modulation scheme employed has a spectral efficiency of 1.68 bps!
Hz. Assuming that the channel impairments necessitate the use of rate 1/2
FEC codes, find the upper bound on the transmission bit rate that a speech
coder used in this system should provide?

Solution to Example 7.4

'Ibtal Bandwidth available for traffic channels = 0.9 x (810—826) = 14.4 MHz.
386 Ch. 7 Speech Coding

Table 7.1 Speech Coders Used In Various Mobile Radio Systems

. Speech Coder Type Used
Standard Service Type .

Bit Rate (kbps)

GSM Cellular RPE-LTP 13

CD-900 Cellular SBC 16
USDC (15-54) Cellular VSELP 8
15-95 Cellular CELP 1.2, 2.4, 4.8, 9.6
15-95 PCS PCS CELP 14.4
PDC Cellular VSELP 4.5, 6.7, 11.2
CT2 Cordless ADPCM 32
DECT Cordless ADPCM 32
PHS Cordless ADPCM 32
DCS-1800 PCS RPE-LTP 13
PACS PCS AJJPCM 32

Number of simultaneous users = 1150.

Therefore, maximum channel bandwidth = 14.4 /1150 MHz = 12.5 kHz.
Spectral Efficiency = 1.68 bps/Hz.
Therefore, maximum channel data rate = 1.68 x 12500 bps = 21kbps.
FEC coder rate = 0.5.
Therefore, maximum net data rate = 21 x 0.5 kbps = 10.5 kbps.
Therefore, we need to design a speech coder with a data rate less than or equal
to 10.5 kbps.

Example 7.5
The output of a speech coder has bits which contribute to signal quality with
varying degree of importance. Encoding is done on blocks of samples of 20 ms
duration (260 bits of coder output). The first 50 of the encoded speech bits (say
type 1) in each block are considered to be the most significant and hence to pro-
tect them from channel errors are appended with 10 CRC bits and convolution.
ally encoded with a rate 112 FEC coder. The next 132 bits (say type 2) are
appended with 5 CRC bits and the tast 78 bits (say type 3) are not error pro-
tected. Compute the gross channel data rate achievable.

Solution to Example 7.5

Number of type 1 channel bits to be transmitted every 20 ms
(5+lO)x2 = 120 bits
Number of type 2 channel bits to be transmitted every 20 ms
132 +5 = 137 bits
Number of type 3 channel bits to be encoded = 78 bits
Total number of channel bits to be transmitted every 20 ms
120 + 137 +78 bits = 335 bits
The OSM Codec 387

Therefore, gross channel bit rate = 335/ (20 x IC3) = 16.75 kbps.

7.9 The GSM Codec

The speech coder used in the pan-European digital cellular standard GSM
goes by a rather grandiose name of regular pulse excited long term prediction
(RPE-LTP) codec. This codec has a net bit rate of 13 kbps and was chosen after
conducting exhaustive subjective tests [Co189] on various competing codecs.
The RPE-LTP codec [Var88} combines the advantages of the earlier French
proposed baseband RELP codec with those of the multi-pulse excited long-term
prediction (MPE-LTP) codec proposed by Germany. The advantage of the base
band RELP codec is that it provides good quality speech at low complexity. The
speech quality of a RELP codec is, however, limited due to the tonal noise intro-
duced by the process of high frequency regeneration and by the bit errors intro-
duced during transmission. The MPE-LTP technique, on the other hand,
produces excellent speech quality at high complexity and is not much affected by
bit errors in the channel. By modifying the RELP codec to incorporate certain
features of the MPE-LTP codec, the net bit rate was reduced from 14.77 kbps to
13.0 kbps without loss of quality. The most important modification was the addi-
tion of a long-term prediction ioop.
The GSM codec is relatively complex and power hungry. Figure 7.10 shows
a block diagram of the speech encoder [Ste94]. The encoder is comprised of four
major processing blocks. The speech sequence is first pre-emphasized, ordered
into segments of 20 ms duration, and then Hamming-windowed. This is followed
by short-term prediction (STP) filtering analysis where the logarithmic area
ratios (L&R's) of the reflection coefficients (k) (8 in number) are computed.
The 8 LAR parameters have different dynamic ranges and probability distribu-
tion functions, and hence all of them are not encoded with the same number of
bits for transmission. The LAR parameters are also decoded by the LPC inverse
filter so as to minimize the error
LTP analysis which involves finding the pitch period and gain factor
is then carried out such that the LTP residual is minimized. To minimize
pitch extraction is done by the LTP by determining that value of delay, D, which
maximizes the crosscorrelation between the current STP error sample, er, and a
previous error sample - D• The extracted pitch and gain are transmitted
and encoded at a rate of 3.6 kbps. The LTP residual. is weighted and decom-
posed into three candidate excitation sequences. The energies of these sequences
are identified, and the one with the highest energy is selected to represent the
LTP residual. The pulses in the excitation sequence are normalized to the high-
est amplitude, quantized, and transmitted at a rate of 9.6 kbps.
Figure 7.11 shows a block diagram of the GSM speech decoder [Ste94]. It
consists of four blocks which perform operations complementary to those of the
388 Ch. 7. Speech Coding

RPE

coded
speech

Figure 710
Block diagram of GSM speech encoder

encoder. The received excitation parameters are RPE decoded arid passed to the
LW synthesis filter which uses the pitch and gain parameter to synthesize the
long-term signal. Short-term synthesis is carried out using the received reflec-
tion coefficients to recreate the original speech signal.

output

Figure 7.11
Block diagram of GSM speech decoder.

Every 260 bits of the coder output (i.e. 20 ms blocks of speech) are ordered,
depending on their importance, into groups of 50, 132, and 78 bits each. The bits
in the first group are very important bits called type [a bits. The next 132 bits
 The USDC Codec 389

areimportant bits called lb bits, and the last 78 bits are called type II bits. Since
type Ia bits are the ones which effect speech quality the most, they have error
detection CRC bits added. Both Ia and lb bits are convolutionally encoded for for-
ward error correction. The least significant type II bits have no error correction
or detection.

7.1D The USDC Codec

The U.S. digital cellular system (IS-54) uses a vector sum excited linear
predictive coder (VSELP). This coder operates at a raw data rate of 7950 bits/s
and a total data rate of• 13 kbps after channel coding. The VSELP coder was
developed by a consortium of companies, and the Motorola implementation was
selected as the speech coding standard after extensive testing.
The VSELP speech coder is a variant of the CELP type vocoders [Ger9OJ.
This coder was designed to accomplish the three goals of highest speech quality,
modest computational complexity, and robustness to channel errors. The code
books in the VSELP encoder are organized with a predefined structure such that
a brute-force search is avoided. This significantly reduces the time required for
the optimum code word search. These code books also impart high speech quality
and increased robustness to channel errors while maintaining modest complex-
ity.
Figure 7.12 shows a block diagram of a VSELP encoder. The 8kbps VSELP
codec utilizes three excitation sources. One is from the long-term ("pitch") predic-
tor state, or adaptive code book. The second and third sources are from the two
VSELP excitation code books. Each of these VSELP code books contain the
equivalent of 128 vectors. These three excitation sequences are multiplied by
their corresponding gain terms and summed to give the combined excitation
sequence. After each subframe the combined excitation sequence is used to
update the long-term filter state (adaptive code book). The synthesis filter is a
direct form 10th order LPC all pole-filter. The LPC coefficients are coded once per
20 ms frame and updated in each 5 ms subframe. The number of samples in a
subframe is 40 at an 8 kHz sampling rate. The decoder is shown in Figure 7.13.

7.11 Performance Evaluation of Speech Coders

There are two approaches to evaluating the performance of a speech coder
in terms of its ability to preserve the signal quality [Jay84]. Objective measures
have the general nature of a signal-to-noise ratio and provide a quantitative
valueS of how well the reconstructed speech approximates the original speech.
Mean square error (MSE) distortion, frequency weighted MSE, and segmented
SNR, articulation index are examples of objective measures. While objective
measures are useful in initial design and simulation of coding systems, they do
not necessarily give an indication of speech quality as perceived by the human
 390 Ch. 7 Speech Coding

Sn

Filter Coefficients

Figure 7.12
Block diagram of the USDC speech encoder.

ear. Since the listener is the ultimate judge of the signal quality, subjective lis-
tening tests constitute an integral part of speech coder evaluation.
Subjective listening tests are conducted by playing the sample to a number
of listeners and asking them to judge the quality of the speech. Speech coders are
highly speaker dependent in that the quality varies with the age and gender of
the speaker, the speed at which the speaker speaks and other factors. The sub-
jective tests are carried out in different environments to simulate real life condi-
tions such as noisy, multiple speakers, etc. These tests provide results in terms of
overall quality, listening effort, intelligibility, and naturalness. The intelligibility
tests measure the listeners ability to identify the spoken word. The diagnostic
rhyme test (OWl') is the most popular and widely used intelligibility test. In this
test a word from a pair of rhymed words such as "those-dose" is presented to the
listener and the listener is asked to identify which word was spoken. Typical per-
centage correct on the DRT tests range from 75-90. The diagnostic acceptability
measure (DAM) is another test that evaluates acceptability of speech coding sys-
tems. All these tests results are difficult to rank and hence require a reference
system. The most popular ranking system is known as the mean opinion score or
MOS ranking. This is a five point quality ranking scale with each point associ-
ated with a standardized descriptions: bad, poor, fair, good, excellent, Table 7.2
gives a listing of the mean square opinion ranking system.
 Pertormance Evakation of Speech Coders 391

7]

I
Output

72

Filter Coefficients

Figure 7.13
Block diagram of the USDC speech decoder.

Table 7.2 lAOS Quality Rating fCoIB9J

Quality Scale Scare Listening Effort Scale

Excellent 5 Na effort required

Good 4 No appreciable effort required
Fair 3 Moderate effort required
Poor 2 Considerable effort required
Bad 1 No meaning understood with reasonable effort

One of the most difficult conditions for speech coders to perform well in is
the case where a digital speech-coded signal is transmitted from the mobile to
the base station, and then demadulated into an analog signal which is then
speech coded for retransmission as a digital signal over a landline or wireless
link. This situation, called tandem signaling, tends to exaggerate the bit errors
originally received at the base station. Tandem signaling is difficult to protect
against, but is an important evaluation criterion in the evaluation of speech cod-
ers. As wireless systems proliferate, there will be a greater demand for mobile-
to-mobile communications, and such links will, by definition, involve at least two
independent, noisy tandems.