The FreeSpeech project (GPL’ed speech recognition) is coming
back to life and looking for volunteers.
Thanks to Stephen Crowley
for the tip.
Jean-Marc.Valin posted
to the Freespeech mailing list:
Hi,
As many of you noticed, there hasn’t been much activity on
SpeechInput for the mast few months. However, a month ago, I
started working again (almost full-time) on the project, which
changed name (once again!) for FreeSpeech.
Also, we are now hosted by on.openprojects.net, the new web page
is at:
http://freespeech.on.openprojects.net/
and we now have anonymous cvs at:
CVSROOT=:pserver:anonymous@cvs.on.openprojects.net:/cvs/freespeech
(simply hit return as the password)
The new mailing list is
freespeech@freespeech.on.openprojects.net and you can subscribe
from http://on.openprojects.net/mailman/listinfo/freespeech
(the SpeechInput@TZoNE.ORG mailing list will no longer be
maintained) For those of you who want to contribute code, I can
create cvs accounts with write access.
Now what’s new: We are now three people working almost full-time
on the project and the total code in the cvs has more than doubled
in the past month. For those of you who didn’t like the front-end
that was there… it’s been replace by the ‘data-flow’ module,
which is far more general (configurable), easier to understand, and
documented with DOC++ comments. It can already generate ceps from a
raw audio file.
In the GMM module, there are now working covariances (diagonal),
gaussians and GMM’s. So those of you who are still interrested in
the project can download the code and contact me if you have any
questions at valj01@gel.usherb.ca
Jean-Marc Valin
P.S. If anyone is interrested in updating the web pages, just
contact me.