Building Speech Synthesis Voices in Festival
In our continuing goal of making speech synthesis more accessible,
this site brings together documentation, scripts programs and examples
of how to build new voices within the framework of Edinburgh
University's Festival Speech
Synthesis System.
The festvox project is run by
Alan W Black of the Language Technologies Institute and
Kevin A. Lenzo of ISRI both
at Carnegie Mellon University.
- festvox documentation and distribution
including scripts and examples for building new speech synthesis voices
in supported and new languages.
New release
A new release of the festvox document, and this website is iminent. The
new release will provide:
- Updates to text and scripts for the existing functions
- Proper configuration and integration with Festival 1.4.1
- Support for building
limited domain synthesizers (often fully
automatically)
- Support for generating basic diphone voice skeleton files
- Integrated support for building new English voices with the
minimal of work (usually just recording and running of scripts from
a walkthrough).