To the best of my knowledge, we haven't run SPRACHcore on Windows
since Win98. SPRACHcore tools were developed on Solaris, and work fine
on Linux. I believe most of the tools run on Mac OS X as well.

Under Windows, your best bet is certainly Cygwin or maybe Mingw (Unix
compatibility libraries for Windows).

Note that SPRACHcore is large and complex, and not for the
faint-hearted. It is also no longer the state-of-the-art in speech
recognition (though it's still rather good at everything other than
"decoding").

Most of the tools have man pages in the distribution. The best way to
see how the tools work together is to look at the simplebn script.
It's a full recognizer, from audio capture to text output.

The definitive reference for the hybrid system is probably the book
"Connectionist Speech Recognition: A Hybrid Approach" by Bourlard and
Morgan. For papers on both hybrid and tandem, check out the ICSI
publications page (towards the bottom):

http://www.icsi.berkeley.edu/cgi-bin/pubs/browse.pl?groupid=000001


Good luck,

Adam Janin
janin@...



francisspeech <francisspeech@...> writes:
> Dear all,
>
> I am a new user of SPRACHcore tools, and have a few questions to ask:
>
> (1) How should I install SPRACHcore under windows XP? It seems that
> there are a lot of packages for the whole tool. Do I need to install
> them one by one? I downloaded the latest version
> (SPRACHcore-2004-08-26), and read the "README" file but couldn't
> figure out how to do it.
>
> (2) By reading through this forum, it seems that some packages have
> been updated to newer versions as compared to the one provided from
> the SPRACHcore download. Can I just replace those packages to use the
> new features?
>
> (3) Is there a tutorial documentation about connectionist ASR approach
> using SPRACHcore tools?
>
> (4) Regarding using Quicknet3 for "hybrid" ASR or "tandem" ASR, are
> there any documentation or tutorials available?
>
> (5) For all of the ICSI speech tools, by any chance, is there a
> distribution of binaries for windows XP?
>
> I would really appreciate your kind suggestions and helps.
> Thanks in advance!
>
> Best regards,
> Frank