Voice Compression and Communications Principles and Applications for Fixed and Wireless Channels

by Hanzo, Lajos; Somerville, F. Clare A.; Woodward, Jason P.

• ISBN13: 9780471150398
• ISBN10: 0471150398
• Edition: 1st
• Format: Hardcover
• Copyright: 2001-09-11
• Publisher: Wiley-IEEE Press

Summary

Up-to-date, expert coverage of topics in wireless voice communications Voice communication is the most important facet of mobile radio service. Even when the predicted surge of wireless data and Internet services becomes a reality, voice will remain the most natural means of human communication. Voice Compression and Communications details issues in wireless voice communications and treats compression, channel coding, and wireless transmission as a joint subject. Part I covers background material, whereas Part II provides detailed information on both proprietary and standardized analysis-by-synthesis codecs, including the speech codecs of virtually all existing wireline-based and wireless systems. Parts III and IV discuss mainly research-based wideband, audio, as well as very low-rate schemes likely to find their way into future standards. Voice Compression and Communications describes fundamental concepts in a non-mathematical way early in the book for those with only a background knowledge of signal processing and communications. More advanced readers will find detailed discussions of theoretical principles, future concepts, and solutions to various specific wireless voice communications problems.

Author Biography

LAJOS HANZO has coauthored five books on mobile radio communications and published more than 300 research papers on a variety of topics in wireless multimedia communications. He holds a chair in telecommunications at the Department of Electronics and Computer Science, University of Southampton, UK, and he is an IEEE Distinguished Lecturer. <p> F. CLARE A. SOMERVILLE is with the Global Wireless Systems Research Department, Bell Laboratories, Swindon, UK. His current research involves real-time techniques for transmission of voice over GPRS and the resultant speech quality attained. <p> P. WOODARD is with UbiNetics Ltd., where he is responsible for the development and implementation of various algorithms for third-generation mobile communications products.

Table of Contents

Preface	xxiii
Acknowledgments	xxix
Part I Speech Signals and Waveform Coding	1	(62)
Speech Signals and Introduction to Speech Coding	3	(24)
Motivation of Speech Compression	3	(1)
Basic Characterization of Speech Signals	4	(3)
Classification of Speech Codecs	7	(4)
Waveform Coding	8	(1)
Time-Domain Waveform Coding	8	(1)
Frequency-Domain Waveform Coding	9	(1)
Vocoders	9	(1)
Hybrid Coding	10	(1)
Waveform Coding	11	(15)
Digitization of Speech	11	(1)
Quantization Characteristics	12	(1)
Quantization Noise and Rate-Distortion Theory	13	(2)
Nonuniform Quantization for a Known PDF: Companding	15	(2)
PDF-Independent Quantization Using Logarithmic Compression	17	(2)
The μ-Law Compander	19	(1)
The A-Law Compander	20	(1)
Optimum Nonuniform Quantization	21	(5)
Chapter Summary	26	(1)
Predicative Coding	27	(36)
Forward Predictive Coding	27	(1)
DPCM Codec Schematic	28	(1)
Predictor Design	29	(7)
Problem Formulation	29	(2)
Covariance Coefficient Computation	31	(1)
Predictor Coefficient Computation	32	(4)
Adaptive One-Word-Memory Quantization	36	(1)
DPCM Performance	37	(2)
Backward-Adaptive Prediction	39	(4)
Background	39	(2)
Stochastic Model Processes	41	(2)
The 32 kbps G.721 ADPCM Codec	43	(6)
Functional Description of the G.721 Codec	43	(1)
Adaptive Quantizer	44	(1)
G.721 Quantizer Scale Factor Adaptation	45	(1)
G.721 Adaptation Speed Control	46	(1)
G.721 Adaptive Prediction and Signal Reconstruction	47	(2)
Subjective and Objective Speech Quality	49	(1)
Variable-Rate G.726 and Embeded G.727 ADPCM	50	(8)
Motivation	50	(1)
Embedded G.727 ADPCM Coding	50	(2)
Performance of the Embedded G.727 ADPCM Codec	52	(6)
Rate-Distortion in Predictive Coding	58	(4)
Chapter Summary	62	(1)
Part II Analysis by Synthesis Coding	63	(276)
Analysis-by-Synthesis Principles	65	(25)
Motivation	65	(1)
Analysis-by-Synthesis Codec Structure	66	(1)
The Short-Term Synthesis Filter	67	(3)
Long-Term Prediction	70	(8)
Open-Loop Optimization of LTP Parameters	70	(5)
Closed-Loop Optimization of LTP Parameters	75	(3)
Excitation Models	78	(3)
Adaptive Short-Term and Long-Term Post-Filtering	81	(2)
Lattice-Based Linear Prediction	83	(6)
Chapter Summary	89	(1)
Speech Spectral Quantization	90	(37)
Log-Area Ratios	90	(5)
Line Spectral Frequencies	95	(10)
Derivation of the Line Spectral Frequencies	95	(4)
Computation of the Line Spectral Frequencies	99	(1)
Chebyshev-Description of Line Spectral Frequencies	100	(5)
Vector Quantization of Spectral Parameters	105	(8)
Background	105	(1)
Speaker-Adaptive Vector Quantization of LSFs	105	(2)
Stochastic VQ of LPC Parameters	107	(1)
Background	107	(2)
The Stochastic VQ Algorithm	109	(2)
Robust Vector Quantization Schemes for LSFs	111	(1)
LSF Vector-Quantizers in Standard Codecs	112	(1)
Spectral Quantizers for Wideband Speech Coding	113	(13)
Introduction to Wideband Spectral Quantization	113	(2)
Statistical Properties of Wideband LSFs	115	(1)
Speech Codec Specifications	116	(1)
Wideband LSF Vector Quantizers	117	(1)
Memoryless Vector Quantization	117	(3)
Predictive Vector Quantization	120	(2)
Multimode Vector Quantization	122	(3)
Simulation Results and Subjective Evaluations	125	(1)
Conclusions on Wideband Spectral Quantization	126	(1)
Chapter Summary	126	(1)
Regular Pulse Excited Coding	127	(18)
Theoretical Background	127	(6)
The 13 kbps RPE-LTP GSM Speech Encoder	133	(4)
Pre-Processing	133	(2)
STP Analysis Filtering	135	(1)
LTP Analysis Filtering	136	(1)
Regular Excitation Pulse Computation	136	(1)
The 13 kbps RPE-LTP GSM Speech Decoder	137	(3)
Bit Sensitivity of the 13 kbps GSM RPE-LTP Codec	140	(2)
Application Example: A Toolbox-Based Speech Transceiver	142	(2)
Chapter Summary	144	(1)
Forward-Adaptive Code Excited Linear Prediction	145	(62)
Background	145	(1)
The Original CELP Approach	146	(3)
Fixed Codebook Search	149	(2)
CELP Excitation Models	151	(9)
Binary-Pulse Excitation	151	(1)
Transformed Binary Pulse Excitation	151	(1)
Excitation Generation	151	(2)
Bit Sensitivity Analysis of the 48 kbps TBPE Speech Codec	153	(2)
Dual-Rate Algebraic CELP Coding	155	(1)
ACELP Codebook Structure	155	(3)
Dual-Rate ACELP Bit Allocation	158	(1)
Dual-Rate ACELP Codec Performance	158	(2)
Optimization of the CELP Codec Parameters	160	(15)
Introduction	160	(1)
Calculation of the Excitation Parameters	160	(1)
Full Codebook Search Theory	160	(2)
Sequential Search Procedure	162	(1)
Full Search Procedure	163	(1)
Suboptimal Search Procedures	164	(1)
Quantization of the Codebook Gains	165	(2)
Calculation of the Synthesis Filter Parameters	167	(1)
Bandwidth Expansion	168	(1)
Least Squares Techniques	168	(3)
Optimization via Powell's Method	171	(1)
Simulated Annealing and the Effects of Quantization	171	(4)
The Error-Sensitivity of CELP Codecs	175	(12)
Introduction	175	(1)
Improving the Spectral Information Error Sensitivity	176	(1)
LSF Ordering Policies	176	(2)
The Effect of FEC on the Spectral Parameters	178	(1)
The Effect of Interpolation	179	(1)
Improving the Error Sensitivity of the Excitation Parameters	180	(1)
The Fixed Codebook Index	180	(1)
The Fixed Codebook Gain	180	(1)
Adaptive Codebook Delay	181	(1)
Adaptive Codebook Gain	182	(1)
Matching Channel Codecs to the Speech Codec	182	(4)
Error-Resilience Conclusions	186	(1)
Application Example: A Dual-Mode 3.1 kBd Speech Transceiver	187	(13)
The Transceiver Scheme	187	(1)
Reconfigurable Modulation	188	(2)
Source-Matched Error Protection	190	(1)
Low-Quality 3.1 kBd Mode	190	(2)
High-Quality 3.1 kBd Mode	192	(3)
Voice Activity Detection and Packet Reservation Multiple Access	195	(1)
kBd System Performance	196	(3)
kBd System Summary	199	(1)
Multi-Slot PRMA Transceiver	200	(6)
Background and Motivation	200	(1)
PRMA-Assisted Multi-Slot Adaptive Modulation	201	(1)
Adaptive GSM-Like Schemes	202	(2)
Adaptive DECT-Like Schemes	204	(1)
Summary of Adaptive Multi-Slot PRMA	205	(1)
Chapter Summary	206	(1)
Standard Forword-Adaptive CELP Codecs	207	(72)
Background	207	(1)
The U.S. DoD FS-1016 4.8 kbits/s CELP Codec	207	(6)
Introduction	207	(2)
LPC Analysis and Quantization	209	(1)
The Adaptive Codebook	210	(1)
The Fixed Codebook	210	(2)
Error Concealment Techniques	212	(1)
Decoder Post-Filtering	212	(1)
Conclusion	212	(1)
The IS-54 DAMPS kbps Pan American Speech Codec	213	(3)
The 6.7 kbps Japanese Digital Cellular System's Speech Codec	216	(2)
The Qualcomm Variable-Rate CELP Codec	218	(7)
Introduction	218	(1)
Codec Schematic and Bit Allocation	219	(1)
Codec Rate Selection	219	(1)
LPC Analysis and Quantization	220	(2)
The Pitch Filter	222	(1)
The Fixed Codebook	223	(1)
Rate 1/8 Filter Excitation	223	(1)
Decoder Post-Filtering	224	(1)
Error Protection and Concealment Techniques	224	(1)
Conclusion	225	(1)
Japanese Half-Rate Speech Codec	225	(8)
Introduction	225	(1)
Codec Schematic and Bit Allocation	226	(2)
Encoder Pre-Processing	228	(1)
LPC Analysis and Quantization	228	(1)
The Weighting Filter	229	(1)
Excitation Vector 1	229	(2)
Excitation Vector 2	231	(1)
Channel Coding	231	(1)
Decoder Post-Processing	232	(1)
The Half-Rate GSM Codec	233	(4)
Half-Rate GSM Codec Outline and Bit Allocation	233	(2)
Spectral Quantization in the Half-Rate GSM Codec	235	(1)
Error Protection	236	(1)
The 8 kbits/s G.729 Codec	237	(19)
Introduction	237	(1)
Codec Schematic and Bit Allocation	237	(1)
Encoder Pre-Processing	238	(1)
LPC Analysis and Quantization	239	(2)
The Weighting Filter	241	(1)
The Adaptive Codebook	242	(1)
The Fixed Algebraic Codebook	242	(3)
Quantization of the Gains	245	(1)
Decoder Post-Processing	246	(2)
G.729 Error Concealment Techniques	248	(1)
G.729 Bit Sensitivity	249	(1)
Orthogonal Frequency Division Multiplex Transmission of G.729 Encoded Speech	250	(1)
Background	250	(1)
System Overview	251	(1)
Turbo Channel Encoding	251	(1)
OFDM in the FRAMES Speech/Data Sub-Burst	252	(1)
Channel Model	253	(1)
Turbo-Coded G.729 OFDM Parameters	253	(1)
Turbo-Coded G.729 OFDM Performance	254	(1)
Turbo-Coded G.729 OFDM Summary	255	(1)
G.729 Summary	256	(1)
The Reduced Complexity G.729 Annex A Codec	256	(3)
Introduction	256	(1)
The Perceptual Weighting Filter	257	(1)
The Open-Loop Pitch Search	257	(1)
The Closed-Loop Pitch Search	257	(1)
The Algebraic Codebook Search	258	(1)
The Decoder Post-Processing	258	(1)
Conclusions	259	(1)
The 12.2 kbps Enhanced Full-Rate GSM Speech Codec	259	(5)
Codec Outline	259	(2)
Enhanced Full-Rate EFR-GSM Encoder	261	(1)
Spectral Quantization in the Enhanced Full-Rate GSM Codec	261	(2)
Adaptive Codebook Search	263	(1)
Fixed Codebook Search	263	(1)
The Enhanced Full-Rate 7.4 kbps IS-136 Speech Codec	264	(4)
IS-136 Codec Outline	264	(1)
IS-136 Bit-Allocation Scheme	265	(2)
Fixed Codebook Search	267	(1)
IS-136 Channel Coding	267	(1)
The ITU G.723.1 Dual-Rate Codec	268	(9)
Introduction	268	(1)
G.723.1 Encoding Principle	269	(1)
Vector-Quantization of the LSPs	270	(1)
Formant-based Weighting Filter	271	(1)
The 6.3 kbps High-Rate G.723.1 Excitation	272	(1)
The 5.3 kbps Low-Rate G.723.1 Excitation	273	(1)
G.723.1 Bit-Allocation	274	(2)
G.723.1 Error Sensitivity	276	(1)
Chapter Summary	277	(2)
Backward-Adaptive Code Excited Linear Prediction	279	(60)
Introduction	279	(1)
Motivation and Background	279	(3)
Backward-Adaptive G.728 Codec Schematic	282	(2)
Backward-Adaptive G.728 Coding Algorithm	284	(14)
G.728 Error Weighting	284	(1)
G.728 Windowing	284	(4)
Codebook Gain Adaption	288	(2)
G.728 Codebook Search	290	(2)
G.728 Excitation Vector Quantization	292	(2)
G.728 Adaptive Post-Filtering	294	(1)
Adaptive Long-Term Post-Filtering	294	(3)
G.728 Adaptive Short-Term Post-Filtering	297	(1)
Complexity and Performance of the G.728 Codec	297	(1)
Reduced-Rate G.728-Like Codec: Variable-Length Excitation Vector	298	(2)
The Effects of Long-Term Prediction	300	(5)
Closed-Loop Codebook Training	305	(4)
Reduced-Rate G.728-Like Codec II: Constant-Length Excitation Vector	309	(1)
Programmable-Rate 8-4 kbps Low-Delay CELP Codec	310	(17)
Motivation	310	(1)
8-4 kbps Codec Improvements Due to Increasing Codebook Sizes	311	(1)
8-4 kbps Codecs-Forward Adaption of the Short-Term Synthesis Filter	311	(2)
Forward Adaption of the Long-Term Predictor	313	(1)
Initial Experiments	313	(3)
Quantization of Jointly Optimized Gains	316	(4)
8-4 kbps Codecs-Voiced/Unvoiced Codebooks	320	(1)
Low-Delay Codecs at 4-8 kbits/s	321	(3)
Low-Delay ACELP Codec	324	(3)
Backward-Adaptive Error Sensitivity Issues	327	(6)
The Error Sensitivity of the G.728 Codec	327	(1)
The Error Sensitivity of Our 4-8 kbits/s Low-Delay Codecs	328	(4)
The Error Sensitivity of Our Low-Delay ACELP Codec	332	(1)
A Low-Delay Multimode Speech Transceiver	333	(5)
Background	333	(1)
8-16 kbps Codec Performance	334	(1)
Transmission Issues	335	(1)
Higher-Quality Mode	335	(2)
Lower-Quality Mode	337	(1)
Speech Transceiver Performance	337	(1)
Chapter Summary	338	(1)
Part III Wideband Coding and Transmission	339	(46)
Wideband Speech Coding	341	(44)
Sub-band-ADPCM Wideband Coding at 64 kbps	341	(16)
Introduction and Specifications	341	(1)
G.722 Codec Outline	342	(4)
Principles of Sub-Band Coding	346	(1)
Quandrature Mirror Filtering	346	(1)
Analysis Filtering	346	(2)
Synthesis Filtering	348	(2)
Practical QMF Design Constraints	350	(5)
G.722 Adaptive Quantization and Prediction	355	(2)
G.722 Coding Performance	357	(1)
Wideband Transform Coding at 32 kbps	357	(3)
Background	357	(1)
Transform-Coding Algorithm	357	(3)
Sub-Band-Split Wideband CELP Codecs	360	(3)
Background	360	(1)
Sub-Band-Based Wideband CELP Coding	360	(1)
Motivation	360	(2)
Low-Band Coding	362	(1)
High-Band Coding	362	(1)
Bit-Allocation Scheme	362	(1)
Fullband Wideband ACELP Coding	363	(5)
Wideband ACELP Excitation	363	(3)
Backward-Adaptive 32 kbps Wideband ACELP	366	(1)
Forward-Adaptive 9.6 kbps Wideband ACELP	366	(2)
A Turbo-Coded Burst-by-Burst Adaptive Wideband Speech Transceiver	368	(16)
Background and Motivation	368	(2)
System Overview	370	(1)
System Parameters	371	(1)
Constant Throughput Adaptive Modulation	372	(1)
Adaptive Wideband Transceiver Performance	373	(1)
Multimode Transceiver Adaptation	374	(1)
Transceiver Mode Switching	375	(3)
The Wideband G.722.1 Codec	378	(1)
Audio Codec Overview	378	(1)
Detailed Description of the Audio Codec	379	(2)
Wideband Adaptive System Performance	381	(1)
Audio Frame Error Results	381	(1)
Audio Segmental SNR Performance and Discussions	382	(2)
G.722.1 Audio Transceiver Summary and Conclusions	384	(1)
Chapter Summary	384	(1)
Part IV Very Low-Rate Coding and Transmission	385	(204)
Overview of Low-Rate Speech Coding	387	(24)
Low-Bitrate Speech Coding	387	(13)
Analysis-by-Synthesis Coding	389	(2)
Speech Coding at 2.4 kbps	391	(1)
Background to 2.4 kbps Speech Coding	392	(1)
Frequency Selective Harmonic Coder	393	(1)
Sinusoidal Transform Coder	394	(1)
Multiband Excitation Coders	395	(1)
Sub-Band Linear Prediction Coder	396	(1)
Mixed Excitation Linear Prediction Coder	396	(1)
Waveform Interpolation Coder	397	(1)
Speech Coding Below 2.4 kbps	398	(2)
Linear Predictive Coding Model	400	(3)
Short-Term Prediction	400	(2)
Long-Term Prediction	402	(1)
Final Analysis-by-Synthesis Model	402	(1)
Speech Quality Measurements	403	(3)
Objective Speech Quality Measures	404	(1)
Subjective Speech Quality Measures	404	(1)
2.4 kbps Selection Process	405	(1)
Speech Database	406	(3)
Chapter Summary	409	(2)
Linear Predictive Vocoder	411	(30)
Overview of a Linear Predictive Vocoder	411	(1)
Line Spectrum Frequencies Quantization	412	(5)
Line Spectrum Frequencies Scalar Quantization	412	(2)
Line Spectrum Frequencies Vector Quantization	414	(3)
Pitch Detection	417	(11)
Voiced-Unvoiced Decision	418	(3)
Oversampled Pitch Detector	421	(1)
Pitch Tracking	422	(3)
Computational Complexity	425	(1)
Integer Pitch Detector	426	(2)
Unvoiced Frames	428	(1)
Voiced Frames	429	(1)
Placement of Excitation Pulses	429	(1)
Pulse Energy	429	(1)
Adaptive Post-Filter	430	(2)
Pulse Dispersion Filter	432	(5)
Pulse Dispersion Principles	432	(2)
Pitch-Independent Glottal Pulse Shaping Filter	434	(1)
Pitch-Dependent Glottal Pulse Shaping Filter	435	(2)
Results for Linear Predictive Vocoder	437	(3)
Chapter Summary	440	(1)
Wavelets and Pitch Detection	441	(20)
Conceptual Introduction to Wavelets	441	(3)
Fourier Theory	441	(1)
Wavelet Theory	442	(1)
Detecting Discontinuities with Wavelets	443	(1)
Introduction to Wavelet Mathematics	444	(5)
Multiresolution Analysis	445	(1)
Polynomial Spline Wavelets	446	(1)
Pyramidal Algorithm	447	(1)
Boundary Effects	448	(1)
Pre-Processing the Wavelet Transform Signal	449	(3)
Spurious Pulses	449	(2)
Normalization	451	(1)
Candidate Glottal Pulses	452	(1)
Voiced-Unvoiced Decision	452	(1)
Wavelet-Based Pitch Detector	453	(7)
Dynamic Programming	454	(3)
Autocorrelation Simplification	457	(3)
Summary and Conclusions	460	(1)
Zinc Function Excitation	461	(40)
Introduction	461	(1)
Overview of Prototype Waveform Interpolation Zinc Function Excitation	462	(4)
Coding Scenarios	462	(1)
U-U-U Encoder Scenario	463	(1)
U-U-V Encoder Scenario	463	(1)
V-U-U Encoder Scenario	464	(1)
U-V-U Encoder Scenario	464	(1)
V-V-V Encoder Scenario	465	(1)
V-U-V Encoder Scenario	465	(1)
U-V-V Encoder Scenario	465	(1)
V-V-U Encoder Scenario	465	(1)
U-V Decoder Scenario	465	(1)
U-U Decoder Scenario	465	(1)
V-U Decoder Scenario	466	(1)
V-V Decoder Scenario	466	(1)
Zinc Function Modeling	466	(4)
Error Minimization	467	(1)
Computational Complexity	468	(1)
Reducing the Complexity of Zinc Function Excitation Optimization	469	(1)
Phases of the Zinc Functions	470	(1)
Pitch Detection	470	(3)
Voiced-Unvoiced Boundaries	471	(1)
Pitch Prototype Selection	471	(2)
Voiced Speech	473	(4)
Energy Scaling	475	(1)
Quantization	476	(1)
Excitation Interpolation Between Prototype Segments	477	(6)
ZFE Interpolation Regions	477	(3)
ZFE Amplitude Parameter Interpolation	480	(1)
ZFE Position Parameter Interpolation	480	(1)
Implicit Signaling of Prototype Zero Crossing	481	(1)
Removal of ZFE Pulse Position Signaling and Interpolation	482	(1)
Pitch Synchronous Interpolation of Line Spectrum Frequencies	482	(1)
ZFE Interpolation Example	482	(1)
Unvoiced Speech	483	(1)
Adaptive Post-Filter	483	(1)
Results for Single Zinc Function Excitation	483	(3)
Error Sensitivity of the 1.9 kbps PWI-ZFE Coder	486	(4)
Parameter Sensitivity of the 1.9 kbps PWI-ZFE coder	486	(1)
Line Spectrum Frequencies	487	(1)
Voiced-Unvoiced Flag	487	(1)
Pitch Period	487	(1)
Excitation Amplitude Parameters	487	(1)
Root Mean Square Energy Parameter	487	(1)
Boundary Shift Parameter	488	(1)
Degradation from Bit Corruption	488	(1)
Error Sensitivity Classes	489	(1)
Multiple Zinc Function Excitation	490	(6)
Encoding Algorithm	490	(2)
Performance of Multiple Zinc Function Excitation	492	(4)
A Sixth-Rate, 3.8 kbps GSM-Like Speech Transceiver	496	(4)
Motivation	496	(1)
The Turbo-Coded Sixth-Rate 3.8 kbps GSM-Like System	496	(1)
Turbo Channel Coding	497	(1)
The Turbo-Coded GMSK Transceiver	498	(2)
System Performance Results	500	(1)
Chapter Summary	500	(1)
Mixed-Multiband Excitation	501	(30)
Introduction	501	(1)
Overview of Mixed-Multiband Excitation	502	(2)
Finite Impulse Response Filter	504	(3)
Mixed-Multiband Excitation Encoder	507	(3)
Voicing Strengths	507	(3)
Mixed-Multiband Excitation Decoder	510	(3)
Adaptive Post-Filter	511	(2)
Computational Complexity	513	(1)
Performance of the Mixed-Multiband Excitation Coder	513	(7)
Performance of a Mixed-Multiband Excitation Linear Predictive Coder	514	(3)
Performance of a Mixed-Multiband Excitation and Zinc Function Prototype Excitation Coder	517	(3)
A Higher Rate 3.85 kbps Mixed-Multiband Excitation Scheme	520	(3)
A 2.35 kbit/s Joint-Detection-Based CDMA Speech Transceiver	523	(7)
Background	523	(1)
The Speech Codec's Bit Allocation	524	(1)
The Speech Codec's Error Sensitivity	525	(1)
Channel Coding	526	(1)
The JD-CDMA Speech System	527	(1)
System Performance	528	(1)
Conclusions on the JD-CDMA Speech Transceiver	529	(1)
Chapter Summary	530	(1)
Sinusoidal Transform Coding Below 4 kbps	531	(34)
Introduction	531	(1)
Sinusoidal Analysis of Speech Signals	532	(2)
Sinusoidal Analysis with Peak-Picking	532	(2)
Sinusoidal Analysis Using Analysis-by-Synthesis	534	(1)
Sinusoidal Synthesis of Speech Signals	534	(2)
Frequency, Amplitude, and Phase Interpolation	534	(1)
Overlap-Add Interpolation	535	(1)
Low-Bitrate Sinusoidal Coders	536	(3)
Increased Frame Length	538	(1)
Incorporating Linear Prediction Analysis	538	(1)
Incorporating Prototype Waveform Interpolation	539	(2)
Encoding the Sinusoidal Frequency Component	541	(2)
Determining the Excitation Components	543	(5)
Peak-Picking of the Residual Spectra	543	(1)
Analysis-by-Synthesis of the Residual Spectrum	543	(2)
Computational Complexity	545	(1)
Reducing the Computational Complexity	545	(3)
Quantizing the Excitation Parameters	548	(8)
Encoding the Sinusoidal Amplitudes	548	(1)
Vector Quantization of the Amplitudes	548	(1)
Interpolation and Decimation	549	(1)
Vector Quantization	550	(1)
Vector Quantization Performance	551	(1)
Scalar Quantization of the Amplitudes	552	(1)
Encoding the Sinusoidal Phases	553	(1)
Vector Quantization of the Phases	553	(1)
Encoding the Phases with a Voiced-Unvoiced Switch	554	(1)
Encoding the Sinusoidal Fourier Coefficients	554	(1)
Equivalent Rectangular Bandwidth Scale	555	(1)
Voiced-Unvoiced Flag	556	(1)
Sinusoidal Transform Decoder	556	(2)
Pitch Synchronous Interpolation	557	(1)
Fourier Coefficient Interpolation	557	(1)
Frequency Interpolation	557	(1)
Computational Complexity	557	(1)
Speech Coder Performance	558	(5)
Chapter Summary	563	(2)
Conclusions on Low-Rate Coding	565	(4)
Overview	565	(1)
Listening Tests	565	(2)
Summary of Very Low-Rate Coding	567	(1)
Further Research	568	(1)
Comparison of Speech Codecs and Transceivers	569	(20)
Background to Speech Quality Evaluation	569	(1)
Objective Speech Quality Measures	570	(7)
Introduction	570	(1)
Signal-to-Noise Ratios	571	(1)
Articulation Index	571	(1)
Cepstral Distance	572	(2)
Example: Computation of Cepstral Coefficients	574	(2)
Logarithmic Likelihood Ratio	576	(1)
Euclidean Distance	577	(1)
Subjective Measures	577	(1)
Quality Tests	577	(1)
Comparison of Subjective and Objective Measures	578	(2)
Background	578	(2)
Intelligibility Tests	580	(1)
Subjective Speech Quality of Various Codecs	580	(2)
Error Sensitivity Comparison of Various Codecs	582	(1)
Objective Speech Performance of Various Transceivers	583	(6)
Appendix A Constructing the Quadratic Spline Wavelets	589	(4)
Appendix B Zinc Function Excitation	593	(4)
Appendix C Probability Density Function for Amplitudes	597	(4)
Bibliography	601	(22)
Index	623	(8)
Author Index	631