TTS voices project

[Jonathan Duddington]

In article <1165326418.31269.540.camel at localhost>,
   Jan Claeys <lists at janc.be> wrote:

> Unfortunately, it seems like eSpeak only (tries to) use OSS.

Text-to-speech is basically a function that translates input data
(text) into output data (sound waveform).  How to play that sound on
the computer is a different problem, and one that varies depending on
the operating environment.

I used the PortAudio library because it claims to be platform
independent.  I believe that version 19 of PortAudio uses ALSA.

I use eSpeak together with the KDE Text to Speech system (KTTS).  In
this case, rather than play the sound data, eSpeak gives it back to
KTTS which plays it, probably through ALSA although I don't know the
details.

I'm not sure how these things work on Ubuntu, but I expect it's a
similar situation.  eSpeak can either pass the speech data back to a
calling program by writing it to a WAV file or its stdout.  I think
SpeechDispatcher does this.

Alternatively, there's a shared library version, libespeak.so, which
provides a function call interface and returns sound data in buffers
using a callback function.

Someone suggested I write a "Gnome Speech Driver" to use the eSpeak
shared library, but that's beyond my capabilities, since I don't run
Gnome and I'm not familiar with the environment that it provides. 
Hopefully someone who knows about these things will do it, and I'll
certainly be pleased to assist.