Multilingual Phone Recognition in Indian Languages (SpringerBriefs in Speech Technology)

✍ Scribed by K.E Manjunath

• Publisher: Springer
• Year: 2021
• Tongue: English
• Leaves: 113
• Category: Library

✦ Synopsis

The book presents current research and developments in multilingual speech recognition. The author presents a Multilingual Phone Recognition System (Multi-PRS), developed using a common multilingual phone-set derived from the International Phonetic Alphabets (IPA) based transcription of six Indian languages - Kannada, Telugu, Bengali, Odia, Urdu, and Assamese. The author shows how the performance of Multi-PRS can be improved using tandem features. The book compares Monolingual Phone Recognition Systems (Mono-PRS) versus Multi-PRS and baseline versus tandem system. Methods are proposed to predict Articulatory Features (AFs) from spectral features using Deep Neural Networks (DNN). Multitask learning is explored to improve the prediction accuracy of AFs. Then, the AFs are explored to improve the performance of Multi-PRS using lattice rescoring method of combination and tandem method of combination. The author goes on to develop and evaluate the Language Identification followed by Monolingual phone recognition (LID-Mono) and common multilingual phone-set based multilingual phone recognition systems.

✦ Table of Contents

Preface
Contents
Acronyms
1 Introduction
1.1 Multilingual Phone Recognition
1.2 Articulatory Features for Multilingual Phone Recognition
1.3 Approaches for Multilingual Phone Recognition
1.4 Code-switched Phone Recognition using Multilingual Phone Recognition Systems
1.5 Objective and Scope of the Work
1.6 Proposed Organization of the Book
References
2 Literature Review
2.1 Introduction
2.2 Prior Work on Multilingual Speech Recognition
2.3 Prior Work on Multilingual Speech Recognition using Articulatory Features
2.4 Prior Work on Code-Switched Speech Recognition using Multilingual Speech Recognition Systems
2.5 Summary
References
3 Development and Analysis of Multilingual Phone RecognitionSystem
3.1 Introduction
3.2 Experimental Setup
3.2.1 Multilingual Speech Corpora
3.2.2 Extraction of Mel-frequency Cepstral Coefficients
3.2.3 Training HMMs and DNNs
3.3 Development of Phone Recognition Systems
3.3.1 Development of Monolingual Phone Recognition Systems
3.3.2 Development of Multilingual Phone Recognition Systems
3.3.3 Development of Tandem Multilingual Phone Recognition Systems
3.4 Performance Evaluation of Phone Recognition Systems
3.4.1 Performance Evaluation of Monolingual Phone Recognition Systems
3.4.2 Performance Evaluation of Multilingual Phone Recognition Systems
3.4.3 Performance Evaluation of Tandem Multilingual Phone Recognition Systems
3.5 Discussion of Results
3.5.1 Analysis and Comparison of the Results
3.5.2 Cross-Lingual Analysis
3.6 Summary
References
4 Prediction of Multilingual Articulatory Features
4.1 Introduction
4.2 Articulatory Features Specification
4.3 Articulatory Feature Predictors (AF-Predictors)
4.3.1 Development of Articulatory Feature Predictors
4.3.2 Oracle Articulatory Features
4.3.3 Performance Evaluation of AF-Predictors
4.4 Performance Improvement of AF-Predictors using Multitask Learning (MTL)
4.5 Summary
References
5 Articulatory Features for Multilingual Phone Recognition
5.1 Introduction
5.2 Proposed Approaches for Multilingual Phone Recognition using Articulatory Features
5.2.1 Development of AF-Based Tandem Multilingual Phone Recognition Systems
5.2.2 Fusion of AFs from Multiple AF Groups
5.3 Multitask Learning Based AFs for Multilingual PhoneRecognition
5.4 Summary
References
6 Applications of Multilingual Phone Recognition in Code-Switched and Non-code-Switched Scenarios
6.1 Introduction
6.2 Experimental Setup
6.2.1 Multilingual Speech Corpora
6.2.2 Code-Switched Test Set
6.2.3 Training Support Vector Machines (SVMs)
6.2.4 Extraction of i-vectors
6.3 Approaches for Multilingual Phone Recognition
6.3.1 LID-switched Monolingual Phone Recognition (LID-Mono) Approach
6.3.1.1 Development of Language Identification (LID) System
6.3.1.2 Development of Monolingual Phone Recognition Systems (Mono-PRS)
6.3.2 Multilingual Phone Recognition using Common Multilingual Phone-set (Multi-PRS) Approach
6.4 Evaluation and Comparison of LID-Mono and Multi-PRS Approaches
6.4.1 Non-Code-Switched Scenario
6.4.2 Code-Switched Scenario
6.5 Summary
References
7 Summary and Conclusion
7.1 Summary of the Book
7.2 Contributions of the Book
7.3 Future Scope of Work
Reference
A Support Vector Machines
B Hidden Markov Models for Speech Recognition
Reference
C Deep Neural Networks for Speech Recognition
C.1 FeedForward Neural Networks
C.2 Training Deep Neural Networks
C.3 Interfacing DNN with HMM (DNN-HMMs)
References
Index