Jian ji yu Yin Ma'erkefu mo xing de Yue yu lian xu yu yin shi bie xi tong

by Chow Ka Fai.Thesis (M.Phil.)--Chinese University of Hong Kong, 1998.Includes bibliographical references (leaves [124-132]).Text in English; abstract also in Chinese.by Chow Ka Fai.Chapter 1 --- INTRODUCTION --- p.1Chapter 1.1 --- Speech Recognition Technology --- p.4Chapter 1.2 --- Automatic Recognition of Cantonese Speech --- p.6Chapter 1.3 --- Objectives of the thesis --- p.8Chapter 1.4 --- Thesis Outline --- p.11Chapter 2 --- FUNDAMENTALS OF HMM BASED RECOGNITION SYSTEM --- p.13Chapter 2.1 --- Introduction --- p.13Chapter 2.2 --- HMM Fundamentals --- p.13Chapter 2.2.1 --- HMM Structure and Behavior --- p.13Chapter 2.2.2 --- HMM-based Speech Modeling --- p.15Chapter 2.2.3 --- Mathematics --- p.18Chapter 2.3 --- hmm Based Speech Recognition System --- p.22Chapter 2.3.1 --- Isolated Speech Recognition --- p.23Chapter 2.3.2 --- Connected Speech Recognition --- p.25Chapter 2.4 --- Algorithms for Finding Hidden State Sequence --- p.28Chapter 2.4.1 --- Forward-backward algorithm --- p.29Chapter 2.4.2 --- Viterbi Decoder Algorithm --- p.31Chapter 2.5 --- Parameter Estimation --- p.32Chapter 2.5.1 --- Basic Ideas for Estimation --- p.32Chapter 2.5.2 --- Single Model Re-estimation Using Best State-Time Alignment (HINIT) --- p.36Chapter 2.5.3 --- Single Model Re-estimation Using Baum- Welch Method (HREST) --- p.39Chapter 2.5.4 --- HMM Embedded Re-estimation (HEREST) --- p.41Chapter 2.6 --- Feature Extraction --- p.42Chapter 2.7 --- Summary --- p.47Chapter 3 --- CANTONESE PHONOLOGY AND LANGUAGE PROPERTIES --- p.48Chapter 3.1 --- Introduction --- p.48Chapter 3.2 --- Cantonese and Chinese Language --- p.48Chapter 3.2.1 --- Chinese Words and Characters --- p.48Chapter 3.2.2 --- The Relationship between Cantonese and Chinese Characters --- p.50Chapter 3.3 --- Basic Syllable structure --- p.51Chapter 3.3.1 --- CVC structure --- p.51Chapter 3.3.2 --- Cantonese Phonemes --- p.52Chapter 3.3.3 --- The Initial-Final structure --- p.55Chapter 3.3.4 --- Cantonese Nine Tone System --- p.57Chapter 3.4 --- Acoustic Properties of Cantonese --- p.58Chapter 3.5 --- Cantonese Phonology for Speech Recognition --- p.60Chapter 3.6 --- Summary --- p.62Chapter 4 --- CANTONESE SPEECH DATABASES --- p.64Chapter 4.1 --- Introduction --- p.64Chapter 4.2 --- The Importance of Speech Data --- p.64Chapter 4.3 --- The Demands of Cantonese Speech Databases --- p.67Chapter 4.4 --- Principles in Cantonese Database Development --- p.67Chapter 4.5 --- Resources and Limitations for Database Designs --- p.69Chapter 4.6 --- Details of Speech Databases --- p.69Chapter 4.6.1 --- Multiple speakers' Speech Database (CUWORD) --- p.70Chapter 4.6.2 --- Single Speaker's Speech Database (MYVOICE) --- p.72Chapter 4.7 --- Difficulties and Solutions in Recording Process --- p.76Chapter 4.8 --- Verification of Phonetic Transcription --- p.78Chapter 4.9 --- Summary --- p.79Chapter 5 --- TRAINING OF AN HMM BASED CANTONESE SPEECH RECOGNITION SYSTEM --- p.80Chapter 5.1 --- Introduction --- p.80Chapter 5.2 --- Objectives of HMM Development --- p.81Chapter 5.3 --- The Design of Initial-Final Models --- p.83Chapter 5.4 --- Initialization of Basic Initial-Final Models --- p.84Chapter 5.4.1 --- The Initialization Training with HEREST --- p.85Chapter 5.4.2 --- Refinement of Initialized Models --- p.88Chapter 5.4.3 --- Evaluation of the Models --- p.90Chapter 5.5 --- Training of Connected Speech Speaker Dependent Models --- p.93Chapter 5.5.1 --- Training Strategy --- p.93Chapter 5.5.2 --- Preliminary Result --- p.94Chapter 5.6 --- Design and Training of Context Dependent Initial Final Models --- p.95Chapter 5.6.1 --- Intra-syllable Context Dependent Units --- p.96Chapter 5.6.2 --- The Inter-syllable Context Dependent Units --- p.97Chapter 5.6.3 --- Model Refinement by Using Mixture Incrementing --- p.98Chapter 5.7 --- Training of Speaker Independent Models --- p.99Chapter 5.8 --- Discussions --- p.100Chapter 5.9 --- Summary --- p.101Chapter 6 --- PERFORMANCE ANALYSIS --- p.102Chapter 6.1 --- Substitution Errors --- p.102Chapter 6.1.1 --- Confusion of Long Vowels and Short Vowels for Initial Stop Consonants --- p.102Chapter 6.1.2 --- Confusion of Nasal Endings --- p.103Chapter 6.1.3 --- Confusion of Final Stop Consonants --- p.104Chapter 6.2 --- Insertion Errors and Deletion Errors --- p.105Chapter 6.3 --- Accuracy of Individual Models --- p.106Chapter 6.4 --- The Impact of Individual Models --- p.107Chapter 6.4.1 --- The Expected Error Rate of Initial Models --- p.110Chapter 6.4.2 --- The Expected Error Rate of Final Models --- p.111Chapter 6.5 --- Suggested Solutions for Error Reduction --- p.113Chapter 6.5.1 --- Duration Constraints --- p.113Chapter 6.5.2 --- The Use of Language Model --- p.113Chapter 6.6 --- Summary --- p.114Chapter 7 --- APPLICATIONS EXAMPLES OF THE HMM RECOGNITION SYSTEM --- p.115Chapter 7.1 --- Introduction --- p.115Chapter 7.2 --- Application 1: A Hong Kong Stock Market Inquiry System --- p.116Chapter 7.3 --- Application 2： A Navigating System for Hong Kong Street Map --- p.117Chapter 7.4 --- Automatic Character-to-Phonetic Conversion --- p.118Chapter 7.5 --- Summary --- p.119Chapter 8 --- CONCLUSIONS AND SUGGESTIONS FOR FURTHER WORK --- p.120Chapter 8.1 --- Conclusions --- p.120Chapter 8.2 --- Suggestions for Future Work --- p.122Chapter 8.2.1 --- Development of Continuous Speech Recognition System --- p.122Chapter 8.2.2 --- Implementation of Statistical Language Models --- p.122Chapter 8.2.3 --- Tones for Continuous Speech --- p.123BIBILOGRAPHYAPPENDI