Self-Learning Speaker Identification A System for Enhanced Speech Recognition

✍ Scribed by Herbig, Tobias;Gerl, Franz;Minker, Wolfgang

• Publisher: Springer Berlin Heidelberg
• Year: 2011
• Tongue: English
• Leaves: 178
• Series: Signals and Communication Technology
• Edition: 1st ed. 2011
• Category: Library

✦ Synopsis

Current speech recognition systems are based on speaker independent speech models and suffer from inter-speaker variations in speech signal characteristics. This work develops an integrated approach for speech and speaker recognition in order to gain space for self-learning opportunities of the system. This work introduces a reliable speaker identification which enables the speech recognizer to create robust speaker dependent models In addition, this book gives a new approach to solve the reverse problem, how to improve speech recognition if speakers can be recognized. The speaker identification enables the speaker adaptation to adapt to different speakers which results in an optimal long-term adaptation.

✦ Table of Contents

Cover......Page 1
Front Matter......Page 2
Speaker Specific Speech Recognition......Page 4
Motivation......Page 12
Overview......Page 14
Speaker Identification......Page 17
Front-End......Page 18
Gaussian Mixture Models......Page 21
Open-Set Speaker Identification......Page 25
Motivation......Page 26
Hidden Markov Models......Page 27
Speaker Identification......Page 31
Speaker Adaptation......Page 33
Motivation......Page 34
Gaussian Mixture Models......Page 35
Maximum A Posteriori......Page 37
Motivation......Page 40
Hidden Markov Models......Page 41
Speaker Adaptation......Page 47
Applications for Speaker Adaptation in Speaker Identification and Speech Recognition......Page 49
Stereo-based Piecewise Linear Compensation for Environments......Page 50
Eigen-Environment......Page 52
Summary......Page 53
Multi-Stage Speaker Identification and Speech Recognition......Page 5
Evaluation Setup......Page 7
Introduction......Page 11
Database......Page 6
Speaker Identification......Page 9
Statistical Modeling of the Likelihood Evolution......Page 3
Speaker Identification......Page 13
Speech Production......Page 15
Evaluation of Joint Speaker Identification and Speech Recognition......Page 16
Evaluation of the Reference Implementation......Page 23
Summary......Page 29
Posterior Probability Computation at Run-Time......Page 8
Closed-Set Speaker Identification......Page 20
Bayesian Information Criterion......Page 28
Motivation and Overview......Page 32
Detection of Unknown Speakers......Page 39
Implementation of an Automated Speech Recognizer......Page 45
Motivation......Page 48
Maximum A Posteriori......Page 51
Maximum Likelihood Linear Regression......Page 54
Eigenvoices......Page 55
Extended Maximum A Posteriori......Page 61
Motivation......Page 63
Stereo-based Piecewise Linear Compensation for Environments......Page 64
Eigen-Environment......Page 66
Summary......Page 67
Audio Signal Segmentation......Page 68
Multi-Stage Speaker Identification and Speech Recognition......Page 72
Phoneme Based Speaker Identification......Page 74
First Conclusion......Page 78
Motivation......Page 80
Algorithm......Page 81
Database......Page 85
Evaluation Setup......Page 86
Results......Page 88
Summary......Page 91
Motivation......Page 92
Joint Speaker Identification and Speech Recognition......Page 94
Speaker Specific Speech Recognition......Page 95
Speaker Identification......Page 100
System Architecture......Page 102
Speaker Identification......Page 104
System Architecture......Page 105
Evaluation......Page 106
Evaluation of Joint Speaker Identification and Speech Recognition......Page 107
Evaluation of the Reference Implementation......Page 114
Summary and Discussion......Page 120
Motivation......Page 123
Posterior Probability Depending on the Training Level......Page 124
Statistical Modeling of the Likelihood Evolution......Page 125
Posterior Probability Computation at Run-Time......Page 130
Closed-Set Speaker Tracking......Page 133
Open-Set Speaker Tracking......Page 136
System Architecture......Page 140
Closed-Set Speaker Identification......Page 142
Open-Set Speaker Identification......Page 147
Summary......Page 151
Summary and Conclusion......Page 152
Outlook......Page 156
Back Matter......Page 160