Building Voices in

the Festival Speech Synthesis System

Processes and issues in building speech synthesis voices

Edition 1.2:beta, for Festival Version 1.4.1

10th July 2000

Full distribution with document sources and related scripts festvox.-1.2-beta.tar.gz
festopt_clunits.tar.gz (updated Festival module for limited domain voices)
Postscript version of this document alone festvox.ps.gz
See http://www.festvox.org for latest versions and examples.

NOTE: this document is incomplete

• 1 Abstract
• 2 Copying
  • 2.1 Acknowledgements
• 3 Introduction
  • 3.1 Text-to-speech process
    • 3.1.1 Text analysis
    • 3.1.2 Linguistic analysis
    • 3.1.3 Waveform generation
  • 3.2 Requirements
    • 3.2.1 Hardware/software requirements
    • 3.2.2 Voice in a new language
    • 3.2.3 Voice in an existing language
  • 3.3 Future
• 4 Speech and Speech Synthesis
• 5 Festival Speech Synthesis System
• 6 Festival's Scheme Programming Language
  • 6.1 Overview
  • 6.2 Data Types
  • 6.3 Functions
    • 6.3.1 Core functions
    • 6.3.2 List functions
    • 6.3.3 Arithmetic functions
    • 6.3.4 I/O functions
    • 6.3.5 String functions
    • 6.3.6 System functions
    • 6.3.7 Utterance Functions
    • 6.3.8 Synthesis Functions
  • 6.4 Debugging and Help
  • 6.5 Adding new C++ functions to Scheme
  • 6.6 Regular Expressions
  • 6.7 Some Examples
• 7 Utterances
  • 7.1 Utterance structure
  • 7.2 Modules
  • 7.3 Utterance access
    • 7.3.1 Features as pathnames
    • 7.3.2 Access idioms
  • 7.4 Utterance building
  • 7.5 Extracting features from utterances
• 8 Diphone databases
  • 8.1 Diphone introduction
  • 8.2 Defining a diphone list
    • 8.2.1 Synthesizing prompts
  • 8.3 Recording the diphones
  • 8.4 Labelling the diphones
  • 8.5 Extracting the pitchmarks
  • 8.6 Building LPC parameters
  • 8.7 Defining a diphone voice
  • 8.8 Checking and correcting diphones
  • 8.9 Diphone check list
  • 8.10 US/UK English Walkthrough
• 9 Unit selection databases
  • 9.1 Cluster unit selection
    • 9.1.1 Collecting databases for unit selection
    • 9.1.2 Preliminaries
    • 9.1.3 Building utterance structures for unit selection
    • 9.1.4 Making cepstrum parameter files
    • 9.1.5 Building the clusters
    • 9.1.6 Defining a voice
    • 9.1.7 Cluster Example
  • 9.2 Diphones from general databases
• 10 Limited domain synthesis
  • 10.1 designing the prompts
  • 10.2 customizing the synthesizer front end
  • 10.3 autolabelling issues
  • 10.4 unit size and type
  • 10.5 using limited domain synthesizers
  • 10.6 Telling the time
    • 10.6.1 Designing the prompts
    • 10.6.2 Recording the prompts
    • 10.6.3 Autolabelling the prompts
    • 10.6.4 Extracting pitchmarks and building LPC coefficients
    • 10.6.5 Building a clunit based synthesizer from the utterances
    • 10.6.6 Testing and tuning
  • 10.7 Making it better
• 11 Building prosodic models
  • 11.1 Phrasing
  • 11.2 Accent/Boundary Assignment
  • 11.3 F0 Generation
    • 11.3.1 F0 by rule
    • 11.3.2 F0 by linear regression
    • 11.3.3 Tilt modelling
  • 11.4 Duration
  • 11.5 Prosody Research
  • 11.6 Prosody Walkthrough
    • 11.6.1 Design database
    • 11.6.2 Setup directory structure
    • 11.6.3 Synthesizing prompts
    • 11.6.4 Recording the prompts
    • 11.6.5 Phonetically label prompts
    • 11.6.6 Extract pitchmarks and F0
    • 11.6.7 Build utterance structures
    • 11.6.8 Duration models
    • 11.6.9 F0 contour models
• 12 Lexicons
  • 12.1 Word pronunciations
  • 12.2 Lexicons and addenda
  • 12.3 Out of vocabulary words
  • 12.4 Letter-to-sound rules by hand
  • 12.5 Building letter-to-sound rules
  • 12.6 Post-lexical rules
• 13 Text analysis
  • 13.1 Token to word rules
  • 13.2 Number pronunciation
    • 13.2.1 Multi-token numbers
    • 13.2.2 Declensions
  • 13.3 Homograph disambiguation
  • 13.4 TTS modes
• 14 General issues.
  • 14.1 Selecting a speaker
  • 14.2 Who owns a voice
  • 14.3 Recording under Unix
  • 14.4 Extracting pitchmarks from waveforms
• 15 Full example
• 16 Evaluation
  • 16.1 Does it work at all?
  • 16.2 Sematically unpredictable sentences
• 17 Resources
  • 17.1 Festival resources
  • 17.2 General speech resources
• 18 English phone lists
  • 18.1 US phoneset
  • 18.2 UK phoneset
• 19 References
• Index

This document was generated on 10 July 2000 using the texi2html translator version 1.52.