Re: [g-a-devel] New to List

[Jason White <jason jasonjgw net>]

Eric Johansson <esj eggo org> wrote:
 
> the only really useful speech recognition environment right now is nuances
> naturally speaking followed closely by Microsoft. obviously the main problem
> is they only run on Windows. 

There is work underway, however, to develop high-quality speech recognition
software for Linux. It's mostly centred on the KDE project at the moment - a
tool called Simon - but there are other, related efforts. Linux Weekly News
published an informative article about this a while ago.

Apparently, work is underway to collect large samples of spoken text that can
be used as training for a high-quality recognizer.

I would suggest contacting the relevant developers to find out how your
project can cooperate with theirs. The end result, if successful, would be
that you could drop the Windows dependency - not necessarily in the near
future, but definitely as soon as the quality of the open recognizers reaches
levels acceptable to you.

I can try to track down the article if you have any problems locating the
right references or the appropriate people to contact.