Part I: Tutorial Overview

• 1 The Fundamentals of HTK
  • 1.1 General Principles of HMMs
  • 1.2 Isolated Word Recognition
  • 1.3 Output Probability Specification
  • 1.4 Baum-Welch Re-Estimation
  • 1.5 Recognition and Viterbi Decoding
  • 1.6 Continuous Speech Recognition
• 2 An Overview of the HTK Toolkit
  • 2.1 HTK Software Architecture
  • 2.2 Generic Properties of a HTK Tool
  • 2.3 The Toolkit
    • 2.3.1 Data Preparation Tools
    • 2.3.2 Training Tools
    • 2.3.3 Recognition Tools
    • 2.3.4 Analysis Tool
  • 2.4 Whats New in Version 2.0?
    • 2.4.1 Whats New in Version 2.1?
• 3 A Tutorial Example of Using HTK
  • 3.1 Data Preparation
    • 3.1.1 Step 1 - the Task Grammar
    • 3.1.2 Step 2 - the Dictionary
    • 3.1.3 Step 3 - Recording the Data
    • 3.1.4 Step 4 - Creating the Transcription Files
    • 3.1.5 Step 5 - Coding the Data
  • 3.2 Creating Monophone HMMs
    • 3.2.1 Step 6 - Creating Flat Start Monophones
    • 3.2.2 Step 7 - Fixing the Silence Models
    • 3.2.3 Step 8 - Realigning the Training Data
  • 3.3 Creating Tied-State Triphones
    • 3.3.1 Step 9 - Making Triphones from Monophones
    • 3.3.2 Step 10 - Making Tied-State Triphones
  • 3.4 Recogniser Evaluation
    • 3.4.1 Step 11 - Recognising the Test Data
  • 3.5 Running the Recogniser Live
  • 3.6 Summary