Sound Capture for Human Machine Interfaces: Practical Aspects of Microphone Array Signal Processing

✍ Scribed by Wolfgang Herbordt

• Publisher: Springer
• Year: 2008
• Tongue: English
• Leaves: 278
• Series: Lecture Notes in Control and Information Sciences
• Edition: 1
• Category: Library

✦ Synopsis

With a continuously increasing desire for natural and comfortable human/machine interaction, the acoustic interface of any terminal for multimedia or telecommunication services is challenged to allow seamless and hands-free audio communication. Sound Capture for Human-Machine Interfaces introduces the practical aspects of microphone array signal processing and presents various combinations of beamforming and acoustic echo cancellation.

✦ Table of Contents

Contents......Page 8
1
Introduction......Page 12
2
Space-Time Signals......Page 16
2.1 Propagating Wave F ields......Page 17
2.2 Spatio-temporal Random Fields......Page 23
2.2.1 Statistical Description of Space-Time Signals......Page 24
2.2.2 Spatio-temporal and Spatio-spectral Correlation Matrice......Page 26
2.3 Summar......Page 34
3
Optimum Linear Filtering......Page 36
3.1.1 Structure of a MIMO Optimum Filter......Page 38
3.1.2 Least-Squares Error (LSE) Optimization......Page 39
3.2.1 System Identi.cation......Page 46
3.2.2 Inverse Modeling......Page 47
3.3 Discussion......Page 49
4
Optimum Beamforming
for Wideband Non-stationary Signals......Page 51
4.1.1 Desired Signal......Page 53
4.1.2 Interference......Page 55
4.1.3 Sensor Noise......Page 56
4.1.4 Sensor Signals......Page 57
4.2.1 Concept of Beamforming......Page 58
4.2.2 Beamformer Res ponse and Interference-Independent Performance Measures......Page 61
4.2.3 Interference-Dependent Performance Measures......Page 66
4.2.4 Spatial Aliasing and Sensor Placement......Page 70
4.3 Data-Independent Beamformer Design......Page 73
4.4 Optimum Data-Dependent Beamformer Designs......Page 76
4.4.1 LSE/MMSE Design......Page 77
4.4.2 Linearly-Constrained Least-Squares Error (LCLSE) and Linearly-Constrained Minimum Variance (LCMV) Design......Page 87
4.4.3 Eigenvector Beamformers......Page 101
4.4.4 Suppression of Correlated Interference......Page 103
4.5 Discussion......Page 104
5
A Practical Audio Acquisition System
Using a Robust GSC (RGSC)......Page 108
5.1 Spatio-temporal Constraints......Page 109
5.2 RGSC as an LCLSE Beamformer with Spatio-temporal Constraints......Page 110
5.2.2 Blocking Matrix......Page 111
5.2.3 Interference Canceller......Page 115
5.3 RGSC in the DTFT Domain......Page 116
5.4.1 Blocking Matrix......Page 119
5.4.2 Interference Canceller......Page 122
5.5 Experimental Results for Stationary Acoustic Conditions......Page 124
5.5.1 Performance Measures in the Context of the RGSC......Page 125
5.5.2 Experimental Setup......Page 126
5.5.3 Interference Rejection of the RGSC......Page 127
5.5.4 Cancellation o f the Desired Signal b y the Blocking Matrix......Page 131
5.6.1 Determination of the Optimum Filters for the Blocking Matrix......Page 136
5.6.2 Determination of the Optimum Filters for the Interference Canceller......Page 138
5.7 Relation to Alternative GSC Realizations......Page 139
5.8 Discussion......Page 140
6
Beamforming Combined with Multi-channel
Acoustic Echo Cancellation......Page 142
6.1.1 Problem Statement......Page 143
6.1.2 Challenges......Page 144
6.2 Combination of Beamforming and Acoustic Echo Cancellation......Page 146
6.2.1 ‘AEC First’......Page 147
6.2.2 ‘Beamformer First’......Page 150
6.3 Integration of Acoustic Echo Cancellation into the GSC......Page 152
6.3.1 AEC After the Quiescent Weight Vector (GSAEC)......Page 154
6.3.2 AEC Combined with the Interference Canceller (GEIC)......Page 161
6.4 Discussion......Page 167
7
Efficient Real-Time Realization
of an Acoustic Human/Machine Front-End......Page 171
7.1 Multi-channel Block-Adaptive Filtering in the Discrete Fourier Transform (DFT) Domain......Page 172
7.1.1 Optimization Criterion......Page 173
7.1.2 Adaptive Algorithm......Page 175
7.2.1 RGSC in the DFT Domain......Page 178
7.2.2 Combination with the AEC......Page 185
7.2.3 Real-Time Algorithm and Computational Complexity......Page 189
7.3.1 Comparison of the DFT-Bin-Wise Adaptation with a Full-Band Adaptation......Page 193
7.3.2 Comparison of the RGSC with a GSC Using a Fixed Blocking Matrix......Page 196
7.3.3 Comparison of the Proposed Adaptation Control with an ‘Ideal’ Adaptation Control......Page 199
7.3.4 Application of the RGSC as a Front-End for an Automatic Speech Recognizer (ASR)......Page 201
7.4 Discussion......Page 210
8
Summary and Conclusions......Page 212
A
Estimation of Signal-to-Interference-Plus-Noise
Ratios (SINRs) Exploiting Non-stationarity......Page 216
A.1.1 Principle......Page 217
A.1.2 Biased SINR Estimation in the DTFT Domain......Page 218
A.1.3 Illustration......Page 219
A.2 Unbiased SINR Estimation Using Spatial Coherence Functions......Page 220
A.3.1 Estimation of the PSDs in the DFT Domain......Page 222
A.3.2 Double-Talk Detection......Page 223
A.3.3 Unbiased Estim ation of the SINR......Page 224
A.3.4 Robustness Improvement......Page 225
A.3.5 Summary of the Algorithm and Experimental Results......Page 226
B
Experimental Setups
and Acoustic Environments......Page 231
B.1 Passenger Cabin of a Car......Page 232
B.2 Multimedia Room......Page 233
C.2 Abbreviations and Acronyms......Page 235
C.3 Mathematical Symbols......Page 236
Symbols......Page 237
References......Page 257