[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: dectalk etc. vs soundblaster (fwd)



> I'm puzzled about something.  Just how high is the barrier to getting
> drivers written to go directly from emacspeak and other text-to-voice
> packages to output for any soundblaster-compatible audio card?

The trick is the TTS software.  The DecTalk Express, for example, is
nothing special; it's a 386 and some other signal processing hardware.
There's even a software version of DecTalk, I believe.  Someone just
needs to write a free TTS program that sounds decent.  rsynth was an
attempt at this, but it's truly awful. (No offense to the authors 
intended, but it's barely useable.)  Most, if not all, of the algorithms
the DecTalk uses should be documented and available somewhere, although
I'm not sure if there are applicable patents that would preclude a free
implementation.

Even something on par with the earlier MITTalk and KLATTalk projects
would be immensely useful.  As research projects here at MIT, I imagine
that they were extensively documented.  I've been meaning to do
some checking on this, but haven't yet had a chance.

--Jered
jered mit edu




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]