Microphone Array Signal Processing (Springer Topics in Signal Processing, 1)

✍ Scribed by Jacob Benesty, Jingdong Chen, Yiteng Huang

• Publisher: Springer
• Year: 2008
• Tongue: English
• Leaves: 352
• Category: Library

✦ Synopsis

In the past few years we have written and edited several books in the area of acousticandspeechsignalprocessing. Thereasonbehindthisendeavoristhat there were almost no books available in the literature when we ?rst started while there was (and still is) a real need to publish manuscripts summarizing the most useful ideas, concepts, results, and state-of-the-art algorithms in this important area of research. According to all the feedback we have received so far, we can say that we were right in doing this. Recently, several other researchers have followed us in this journey and have published interesting books with their own visions and perspectives. The idea of writing a book on Microphone Array Signal Processing comes from discussions we have had with many colleagues and friends. As a c- sequence of these discussions, we came up with the conclusion that, again, there is an urgent need for a monograph that carefully explains the theory and implementation of microphone arrays. While there are many manuscripts on antenna arrays from a narrowband perspective (narrowband signals and narrowband processing), the literature is quite scarce when it comes to s- sor arrays explained from a truly broadband perspective. Many algorithms for speech applications were simply borrowed from narrowband antenna - rays. However, a direct application of narrowband ideas to broadband speech processing may not be necessarily appropriate and can lead to many m- understandings.

✦ Table of Contents

Title Page
Preface
Contents
List of Contributors
Linear System Identification in the Short-Time Fourier Transform Domain
Introduction
Problem Formulation
System Identification Using Crossband Filters
Crossband Filters Representation
Batch Estimation of Crossband Filters
Selecting the Optimal Number of Crossband Filters
System Identification Using the MTF Approximation
The MTF Approximation
Optimal Window Length
The Cross-MTF Approximation
Adaptive Estimation of Cross-Terms
Adaptive Control Algorithm
Experimental Results
Crossband Filters Estimation
Comparison of the Crossband Filters and MTF Approaches
CMTF Adaptation for Acoustic Echo Cancellation
Conclusions
Appendix
References
Identification of the Relative Transfer Function between Sensors in the Short-Time Fourier Transform Domain
Introduction
Identification of the RTF Using Multiplicative Transfer Function Approximation
Problem Formulation and the Multiplicative Transfer Function Approximation
RTF Identification Using Non-Stationarity
RTF Identification Using Speech Signals
Identification of the RTF Using Convolutive Transfer Function Approximation
The Convolutive Transfer Function Approximation
RTF Identification Using the Convolutive Transfer Function Approximation
Relative Transfer Function Identification in Speech Enhancement Applications
Blocking Matrix
The Transfer Function Generalized Sidelobe Canceler
Conclusions
References
Representation and Identification of Nonlinear Systems in the Short-Time Fourier Transform Domain
Introduction
Volterra System Identification
Representation of Volterra Filters in the STFT Domain
Second-Order Volterra Filters
High-Order Volterra Filters
A New STFT Model For Nonlinear Systems
Quadratically Nonlinear Model
High-Order Nonlinear Models
Quadratically Nonlinear System Identification
Batch Estimation Scheme
Adaptive Estimation Scheme
Experimental Results
Performance Evaluation for White Gaussian Inputs
Nonlinear Undermodeling in Adaptive System Identification
Nonlinear Acoustic Echo Cancellation Application
Conclusions
Appendix
References
Variable Step-Size Adaptive Filters for Echo Cancellation
Introduction
Non-Parametric VSS-NLMS Algorithm
VSS-NLMS Algorithms for Echo Cancellation
VSS-APA for Echo Cancellation
VFF-RLS for System Identification
Simulations
VSS-NLMS Algorithms for AEC
VSS-APA for AEC
VFF-RLS for System Identification
Conclusions
References
Simultaneous Detection and Estimation Approach for Speech Enhancement and Interference Suppression
Introduction
Classical Speech Enhancement in Nonstationary Noise Environments
Simultaneous Detection and Estimation for Speech Enhancement
Quadratic Distortion Measure
Quadratic Spectral Amplitude Distortion Measure
Spectral Estimation Under a Transient Noise Indication
A Priori SNR Estimation
Experimental Results
Simultaneous Detection and Estimation
Spectral Estimation Under a Transient Noise Indication
Conclusions
References
Speech Dereverberation and Denoising Based on Time Varying Speech Model and Autoregressive Reverberation Model
Introduction
Goal
Technological Background
Minimum Mean-Squared Error Signal Estimation and Model-Based Approach
Dereverberation Method
Heuristic Derivation of Weighted Prediction Error Method
Reverberation Model
Clean Speech Model
Clean Speech Signal Estimator and Parameter Optimization
Combined Dereverberation and Denoising Method
Room Acoustics Model
Clean Speech Model
Clean Speech Signal Estimator
Parameter Optimization
Experiments
Conclusions
References
Codebook Approaches for Single Sensor Speech/Music Separation
Introduction
Single Sensor Source Separation
Problem Formulation
GSMM-Based Source Separation
AR-Based Source Separation
Bayesian Non-Negative Matrix Factorization
Learning the Codebook
Multi-Window Source Separation
General Description of the Algorithm
Choice of a Confidence Measure
Practical Choice of the Thresholds
Estimation of the Expansion Coefficients
Median Filter
Smoothing Prior
GMM Modeling of the Amplitude Coefficients
Experimental Study
Evaluation Criteria
Experimental Setup and Results
Conclusions
References
Microphone Arrays: Fundamental Concepts
Introduction
Signal Model
Array Model
Signal-to-Noise Ratio
Array Gain
Noise Rejection and Desired Signal Cancellation
Beampattern
Anechoic Plane Wave Model
Directivity
Superdirective Beamforming
White Noise Gain
Spatial Aliasing
Monochromatic Signal
Broadband Signal
Mean-Squared Error
Wiener Filter
Minimum Variance Distortionless Response
Conclusions
References
The MVDR Beamformer for Speech Enhancement
Introduction
Problem Formulation
From Speech Distortion Weighted Multichannel Wiener Filter to Minimum Variance Distortionless Response Filter
Speech Distortion Weighted Multichannel Wiener Filter
Minimum Variance Distortionless Response Filter
Decomposition of the Speech Distortion Weighted Multichannel Wiener Filter
Equivalence of MVDR and Maximum SNR Beamformer
Performance Measures
Performance Analysis
On the Comparison of Different MVDR Beamformers
Local Analyzes
Global Analyzes
Non-Coherent Noise Field
Coherent plus Non-Coherent Noise Field
Performance Evaluation
Influence of the Number of Microphones
Influence of the Reverberation Time
Influence of the Noise Field
Example Using Speech Signals
Conclusions
Appendix
References
Extraction of Desired Speech Signals in Multiple-Speaker Reverberant Noisy Environments
Introduction
Problem Formulation
Proposed Method
The LCMV and MVDR Beamformers
The Constraints Set
Equivalent Constraints Set
Modified Constraints Set
Estimation of the Constraints Matrix
Interferences Subspace Estimation
Desired Sources RTF Estimation
Algorithm Summary
Experimental Study
The Test Scenario
Simulated Environment
Real Environment
Conclusions
References
Spherical Microphone Array Beamforming
Introduction
Spherical Array Processing
Regular Beam Pattern
Delay-and-Sum Beam Pattern
Dolph-Chebyshev Beam Pattern
Optimal Beamforming
Beam Pattern with Desired Multiple Nulls
2D Beam Pattern and its Steering
Near-Field Beamforming
Direction-of-Arrival Estimation
Conclusions
References
Steered Beamforming Approaches for Acoustic Source Localization
Introduction
Signal Model
Spatial and Spatiotemporal Filtering
Parameterized Spatial Correlation Matrix (PSCM)
Source Localization Using Parameterized Spatial Correlation
Steered Response Power
Minimum Variance Distortionless Response
Maximum Eigenvalue
Broadband MUSIC
Minimum Entropy
Sparse Representation of the PSCM
Linearly Constrained Minimum Variance
Autoregressive Modeling
Challenges
Conclusions
References
Index