Kyle <speakup at braille.uwo.ca> wrote: >While I agree regarding the standard FestVox voices that come with Festival, >it does offer a framework that allows much more understandable voices to be >installed and used instead. Also, I find eSpeak to be of much higher quality, >easier to understand, available at a much lower price and much better >maintained than IBMTTS, ViaVoice, TTSynth, Voxin or whatever they're calling >that old broken down Jaws voice these days, and that's completely putting >aside for the moment that both Festival and eSpeak are completely free and >open source, meaning that you have the freedom to fix what you don't like >about them. Pico is also open source, although from what I understand, there >is a bit of a problem making new voice data or something. The sources of the voice files are available but the tools to convert them to binary files exist only as MS-Windows executables. these don't include the machine learning algorithms needed to build new voices. there is a need for better-quality free/opens-ource speech synthesis software. Either someone will write it, or someone will release a previously proprietary system under an open licence.
