Fundamentals of Speech Enhancement (eBook)

Jacob Benesty (Autor)

eBook Download: PDF

2018 | 1st ed. 2018
VIII, 106 Seiten
Springer International Publishing (Verlag)
978-3-319-74524-4 (ISBN)

Infos

This book presents and develops several important concepts of speech enhancement in a simple but rigorous way. Many of the ideas are new; not only do they shed light on this old problem but they also offer valuable tips on how to improve on some well-known conventional approaches. The book unifies all aspects of speech enhancement, from single channel, multichannel, beamforming, time domain, frequency domain and time-frequency domain, to binaural in a clear and flexible framework. It starts with an exhaustive discussion on the fundamental best (linear and nonlinear) estimators, showing how they are connected to various important measures such as the coefficient of determination, the correlation coefficient, the conditional correlation coefficient, and the signal-to-noise ratio (SNR). It then goes on to show how to exploit these measures in order to derive all kinds of noise reduction algorithms that can offer an accurate and versatile compromise between noise reduction and speech distortion.

Jacob Benesty received his Master's degree in Microwaves from Pierre & Marie Curie University, France, in 1987, and his Ph.D. in Control and Signal Processing from Orsay University, France, in 1991. From January 1994 to July 1995, he worked at Telecom Paris University, France, on multichannel adaptive filters and acoustic echo cancellation. From October 1995 to May 2003, he was first a consultant and then a member of the technical staff at Bell Laboratories, Murray Hill, New Jersey (NJ), USA. In May 2003, he joined the University of Quebec, INRS-EMT, in Montreal, Canada, as a professor. His research interests include signal processing, acoustic signal processing, and multimedia communications. Dr Benesty received the 2001 Best Paper Award from the IEEE Signal Processing Society. He has co-authored multiple books with Springer and is the series editor for the 'Springer Topics in Signal Processing'.

Inhalt

Abstract 6
Contents 7
1 Introduction 9
1.1 General Formulation of the Speech Enhancement Problem 9
1.2 Organization of the Work 10
References 11
2 Best Speech Enhancement Estimator in the Frequency Domain 12
2.1 Signal Model and Problem Formulation 12
2.2 Laws of Total Expectation and Total Variance 13
2.3 Best Estimator 14
2.4 Example with Gamma Distributions 18
2.4.1 Reformulation of the Problem and Approximation 18
2.4.2 Best Estimator 19
2.5 A Brief Study of the Best Quadratic Estimator 22
2.6 Generalization to the Multichannel Case 24
References 29
3 Best Speech Enhancement Estimator in the Time Domain 30
3.1 Signal Model and Problem Formulation 30
3.2 Best Estimator 32
3.3 Best Linear Estimator 38
3.4 Generalization to the Binaural Case 41
3.4.1 Problem Formulation 41
3.4.2 Best Estimator 44
3.4.3 Best Widely Linear Estimator 49
References 50
4 Speech Enhancement Via Correlation Coefficients 52
4.1 Signal Model and Problem Formulation 52
4.2 Linear Filtering and Correlation Coefficients 53
4.3 Optimal Filters 56
4.3.1 SPCC Between Filter Output and Desired Signal 56
4.3.2 SPCC Between Filter Output and Noise Signal 62
4.3.3 SPCC Between Filter Output and Filtered Desired Signal 66
4.3.4 Other Possibilities 70
References 70
5 On the Output SNR in Speech Enhancement and Beamforming 72
5.1 Signal Model and Problem Formulation 72
5.2 Linear Filtering, Output and Fullmode Input SNRs 73
5.3 Optimal Filters 79
5.3.1 Rank-One Speech Covariance Matrix 79
5.3.2 Rank-Deficient Speech Covariance Matrix 80
5.3.3 Full-Rank Speech Covariance Matrix 82
5.4 Application to Fixed and Superdirective Beamforming 84
References 88
6 Speech Enhancement from the Fullband Output SNR Perspective 89
6.1 Signal Model and Problem Formulation 89
6.2 Speech Enhancement with Gains 90
6.3 Determination of the Optimal Gains 93
6.3.1 Maximization of the Fullband Output SNR 93
6.3.2 Minimization of the Fullband Output SNR 96
6.4 Taking the Interframe Correlation Into Account 98
6.5 Generalization to the Multichannel Case 104
References 110
Index 111

Produkt-Details

• Erscheint lt. Verlag: 9.2.2018
• Reihe/Serie: SpringerBriefs in Electrical and Computer Engineering
• Zusatzinfo: VIII, 106 p.
• Verlagsort: Cham
• Sprache: englisch
• Themenwelt: Technik ► Elektrotechnik / Energietechnik
• Schlagworte: Beamforming • best (linear and nonlinear) estimators • Binaural • coefficient of determination • conditional correlation coefficient • correlation coefficient • noise reduction • Optimal Filters • snr • Speech Enhancement
• ISBN-10: 3-319-74524-7 / 3319745247
• ISBN-13: 978-3-319-74524-4 / 9783319745244