Newsgroups: comp.ai,comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!uunet!psinntp!isc-newsserver!ultb!adw3345
From: adw3345@ultb.isc.rit.edu (A.D. Williams)
Subject: VOICE RECOGNITION
Message-ID: <1995Jan22.165450.22137@ultb.isc.rit.edu>
Sender: news@ultb.isc.rit.edu (USENET News System)
Nntp-Posting-Host: ultb-gw.isc.rit.edu
Reply-To: adw3345@ultb.rit.edu ()
Organization: Rochester Institute of Technology
Date: Sun, 22 Jan 1995 16:54:50 GMT
Lines: 49
Xref: glinda.oz.cs.cmu.edu comp.ai:26708 comp.ai.nat-lang:2700


 Hello!

 I'm not sure what the appopriate newsgroup would be, but I am
interested in finding a discussion about the voice recognition packages
out in the market today and the future of voice recognition.

 My interest stems from an application I would like to develop for some
students in a local high school for the deaf to help with speech 
comprehension.

 Many hard of hearing people can hear sounds, but need some training to
make sense of the syllables and convert them to words. For example, my
application would have someone speak into the microphone and the
phonetic representation would appear on the screen.

 Obviously, it would be important to have the system speaker
independent. Since the goal is not to have a dictation system that
prints out the words being said (that would be self defeating if one
were to see the words "cough", "plough", "tough" without the different
phonetic representations), I don't think this is too much to ask from
the technology available today.

 Currently, I'm favoring IBM's ICSS package running under OS/2 on a
Pentium platform. ICSS has the ability to recognize up to 1000 words
with speaker independence. ICSS uses a BNF tree, and while I'm only
interested in the 48 phonemes, it might be useful to have 1000 commonly
used syllables print out instead of an phonetic alphabet.

 One use for this would be to aid deaf people in using the telephone.
Since there is little input other than sounds (such as lipreading or
guestures) a computer recognizing the phonemes baing spoken would be a
big help.

 Has anyone done something like this? What are your impressions about
ICSS, or any other speech recognition product for that matter? Are there
any journals or books that might be helpful about this subject?

 Sorry to post here. I think it's time for comp.app.voice-recogntion!

                                                Derrick



-- 
-------------------------------------------------------------------------
- Derrick Williams   Rochester Institute of Technology | Insert snappy  -
- adw3345@ultb.isc.rit.edu     Computer Science        | quotation here -
-------------------------------------------------------------------------
