A Hybrid System with Symbolic AI and Statistical Methods for Speech Recognition

By Jesus Savage-Carmona

This document is also available in postscript.

Table of Contents

• Title Page
• Abstract
• List of Figures
• List of Tables
• Acknowledgements
• Chapter 1: Introduction
• Chapter 2:Digital Signal Processing and Statistical Models
  • 2.1 Vector Quantization
  • 2.2 Cepstral Coefficients
  • 2.3 Distortion Measures
  • 2.4 Hidden Markov Models
    • 2.4.1 Discrete Hidden Markov Models For Words
    • 2.4.2 Continuous Hidden Markov Models
  • 2.5 Voice Recognition Using Vector Quantizers
    • 2.5.1 Voice Recognition Using LPC Vector Quantizers
    • 2.5.2 Voice Recognition Using Multi-Section LPC Vector Quantizers
    • 2.5.3 Voice Recognition Using Discrete Hidden Markov Models
  • 2.6 Isolated Voice Recognition Using MSVQ and HMMs
    • 2.6.1 Hardware and Software Implementation
    • 2.6.2 Experiments and Results
• Chapter 3: Keyword Speech Recognition
  • 3.1 Connected Word Recognition
  • 3.2 Recognition of Keywords in Continuous Speech
• Chapter 4: Natural Language Understanding
  • 4.1 Syntactic Analysis
  • 4.2 Semantic Analysis
  • 4.3 Conceptual Dependency
  • 4.4 Expert Systems
    • 4.4.1 CLIPS
  • 4.5 Connected Word Recognition Using Keywords
    • 4.5.1 Keyword Recognition and Conceptual Dependency
    • 4.5.2 Experiments and Results
• Chapter 5: Speech Recognition Applications
  • 5.1 Virtual Reality
  • 5.2 Speech Recognition in a VR system
• Chapter 6: Context Representation
  • 6.1 Context Representation Using Objects
  • 6.2 Context Representation Using Scripts
  • 6.3 The use of Pronouns
• Chapter 7: Planning
  • 7.1 Planning
  • 7.2 Experiments and Results
• Chapter 8: Conclusions and Future Research
  • 8.1 Conclusions
  • 8.2 Contribution of This Dissertation
  • 8.3 Future Research
• Bibliography
• Appendix A
• Appendix B