Microphone Array Signal Processing (eBook)

Preface 6
Contents 8
1 Introduction 12
1.1 Microphone Array Signal Processing 12
1.2 Organization of the Book 16
2 Classical Optimal Filtering 18
2.1 Introduction 18
2.2 Wiener Filter 19
2.3 Frost Filter 27
2.4 Kalman Filter 32
2.5 A Viable Alternative to the MSE 36
2.6 Conclusions 48
3 Conventional Beamforming Techniques 50
3.1 Introduction 50
3.2 Problem Description 51
3.3 Delay-and-Sum Technique 52
3.4 Design of a Fixed Beamformer 57
3.5 Maximum Signal-to-Noise Ratio Filter 60
3.6 Minimum Variance Distortionless Response Filter 63
3.7 Approach with a Reference Signal 65
3.8 Response-Invariant Broadband Beamformers 66
3.9 Null-Steering Technique 69
3.10 Microphone Array Pattern Function 72
3.11 Conclusions 76
4 On the Use of the LCMV Filter in Room Acoustic Environments 78
4.1 Introduction 78
4.2 Signal Models 78
4.3 The LCMV Filter with the Anechoic Model 80
4.4 The LCMV Filter with the Reverberant Model 84
4.5 The LCMV Filter with the Spatio-Temporal Model 86
4.6 The LCMV Filter in the Frequency Domain 92
4.7 Conclusions 94
5 Noise Reduction with Multiple Microphones: a Unified Treatment 96
5.1 Introduction 96
5.2 Signal Model and Problem Description 97
5.3 Some Useful Definitions 98
5.4 Wiener Filter 100
5.5 Subspace Method 103
5.6 Spatio-Temporal Prediction Approach 106
5.7 Case of Perfectly Coherent Noise 108
5.8 Adaptive Noise Cancellation 110
5.9 Kalman Filter 111
5.10 Simulations 112
5.11 Conclusions 125
6 Noncausal (Frequency-Domain) Optimal Filters 126
6.1 Introduction 126
6.2 Signal Model and Problem Formulation 127
6.3 Performance Measures 128
6.4 Noncausal Wiener Filter 131
6.5 Parametric Wiener Filtering 135
6.6 Generalization to the Multichannel Case 137
6.7 Conclusions 147
7 Microphone Arrays from a MIMO Perspective 150
7.1 Introduction 150
7.2 Signal Models and Problem Description 151
7.3 Two-Element Microphone Array 155
7.4 N-Element Microphone Array 161
7.5 Simulations 167
7.6 Conclusions 174
8 Sequential Separation and Dereverberation: the Two- Stage Approach 176
8.1 Introduction 176
8.2 Signal Model and Problem Description 176
8.3 Source Separation 179
8.4 Speech Dereverberation 186
8.5 Conclusions 191
9 Direction-of-Arrival and Time- Difference- of- Arrival Estimation 192
9.1 Introduction 192
9.2 Problem Formulation and Signal Models 195
9.3 Cross-Correlation Method 199
9.4 The Family of the Generalized Cross-Correlation Methods 201
9.5 Spatial Linear Prediction Method 204
9.6 Multichannel Cross-Correlation Coeficient Algorithm 207
9.7 Eigenvector-Based Techniques 211
9.8 Minimum Entropy Method 216
9.9 Adaptive Eigenvalue Decomposition Algorithm 218
9.10 Adaptive Blind Multichannel Identification Based Methods 220
9.11 TDOA Estimation of Multiple Sources 222
9.12 Conclusions 226
10 Unaddressed Problems 228
10.1 Introduction 228
10.2 Speech Source Number Estimation 228
10.3 Cocktail Party Effect and Blind Source Separation 229
10.4 Blind MIMO Identification 231
10.5 Conclusions 233
References 234
Index 248