This book discusses the contribution of excitation source information in discriminating language. The authors focus on the excitation source component of speech for enhancement of language identification (LID) performance. Language specific features are extracted using two different modes: (i) Implicit processing of linear prediction (LP) residual and (ii) Explicit parameterization of linear prediction residual. The book discusses how in implicit processing approach, excitation source features are derived from LP residual, Hilbert envelope (magnitude) of LP residual and Phase of LP residual; and in explicit parameterization approach, LP residual signal is processed in spectral domain to extract the relevant language specific features. The authors further extract source features from these modes, which are combined for enhancing the performance of LID systems. The proposed excitation source features are also investigated for LID in background noisy environments. Each chapter of this book provides the motivation for exploring the specific feature for LID task, and subsequently discuss the methods to extract those features and finally suggest appropriate models to capture the language specific knowledge from the proposed features. Finally, the book discuss about various combinations of spectral and source features, and the desired models to enhance the performance of LID systems. Preface 6 Contents 8 Acronyms 11 1 Introduction 13 1.1 Introduction 13 1.2 Types of Language Identification Systems 14 1.2.1 Explicit Language Identification System 14 1.2.2 Implicit Language Identification System 15 1.3 Features Used for Developing Speech Systems 15 1.4 Issues in Developing Language Identification Systems 17 1.5 Objective and Scope of the Work 18 1.6 Contributions of the Book 18 1.7 Organization of the Book 19 References 20 2 Language Identification---A Brief Review 22 2.1 Prior Works on Explicit Language Identification System 22 2.2 Prior Works on Implicit Language Identification System 28 2.3 Prior Works on Excitation Source Features 32 2.4 Motivation for the Present Work 34 2.5 Summary 38 References 38 3 Implicit Excitation Source Features for Language Identification 42 3.1 Introduction 42 3.2 Speech Corpus 43 3.2.1 Indian Institute of Technology Kharagpur Multi-Lingual Indian Language Speech Corpus (IITKGP-MLILSC) 43 3.2.2 Oregon Graduate Institute Multi-Language Telephone-Based Speech (OGI-MLTS) Database 45 3.3 Extraction of Implicit Excitation Source Information from Linear Prediction Residual 45 3.3.1 Analytic Signal Representation of Linear Prediction Residual 46 3.3.2 Implicit Processing of Linear Prediction Residual Signal 47 3.3.3 Implicit Processing of Magnitude and Phase Components of Linear Prediction Residual 50 3.4 Development of Language Identification Systems Using Implicit Excitation Source Features 50 3.5 Performance Evaluation of LID Systems Developed Using Implicit Excitation Source Features 53 3.6 Evaluation of LID Systems Developed Using Implicit Excitation Source Features on OGI-MLTS Database 61 3.7 Summary 61 References 62 4 Parametric Excitation Source Features for Language Identification 63 4.1 Introduction 63 4.2 Parametric Representation of Excitation Source Information 64 4.2.1 Parametric Representation of Sub-segmental Level Excitation Source Information 64 4.2.2 Parametric Representation of Segmental Level Excitation Source Information 71 4.2.3 Parametric Representation of Supra-Segmental Level Excitation Source Information 74 4.3 Development of LID Systems Using Parametric Features of Excitation Source 76 4.4 Performance Evaluation of LID Systems Developed Using Parametric Features of Excitation Source 78 4.5 Performance Evaluation of LID Systems Developed Using Parametric Features of Excitation Source on OGI-MLTS Database 84 4.6 Summary 84 References 84 5 Complementary and Robust Nature of Excitation Source Features for Language Identification 86 5.1 Introduction 86 5.2 Vocal Tract Features 87 5.3 Development of Language Identification Systems Using Excitation Source and Vocal Tract Features 88 5.4 Performance Evaluation of Source and System Integrated LID Systems 90 5.5 Performance Evaluation of Source and System Integrated LID Systems 95 5.6 Robustness of Excitation Source Features 96 5.6.1 Motivation for the Use of Excitation Source Information for Robust Language Identification 96 5.6.2 Processing of Robust Excitation Source Features for Language Identification 98 5.6.3 Evaluation of Robustness of Excitation Source Features for Language Identification 99 5.7 Summary 105 References 105 6 Summary and Conclusion 106 6.1 Summary of the Book 106 6.2 Contributions of the Book 108 6.3 Future Scope of Work 108 References 109 Appendix AGaussian Mixture Model 110 Appendix BMel-Frequency Cepstral Coefficient (MFCC)Features 114 Appendix CEvaluation of Excitation Source Featuresin Different Noisy Conditions 118 Front Matter....Pages i-xii Introduction....Pages 1-9 Language Identification—A Brief Review....Pages 11-30 Implicit Excitation Source Features for Language Identification....Pages 31-51 Parametric Excitation Source Features for Language Identification....Pages 53-75 Complementary and Robust Nature of Excitation Source Features for Language Identification....Pages 77-96 Summary and Conclusion....Pages 97-100 Back Matter....Pages 101-119
