Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!yale!mintaka!oliveb!bunker!hcap!hnews!129!26!Jack.O'keeffe
From: Jack.O'keeffe@f26.n129.z1.fidonet.org (Jack O'keeffe)
Newsgroups: misc.handicap
Subject: Re: Handicapped Tools
Message-ID: <9508@bunker.UUCP>
Date: 10 Jan 90 04:54:44 GMT
Sender: news@bunker.UUCP
Reply-To: Jack.O'keeffe@f26.n129.z1.fidonet.org
Distribution: misc
Organization: FidoNet node 1:129/26 - SoundingBoard, Pittsburgh PA
Lines: 38
Approved: wtm@bunker.UUCP

Index Number: 6177

 SC> From: SRCLARK@COLOSPGS.BITNET (Steve Clark)
 SC> Message-ID: <9347@bunker.UUCP>

 SC> Index Number: 6031

 SC> I'm an electrical engineer studying at the University of
 SC> Colorado at Colorado Springs. I'm trying to research
 SC> handicapped tools that are available and that are needed.

Steve, if you'd like a really challenging project that would fill
a void and help - literally - millions, try your hand at a large
vocabulary, speaker independent, speech-to-text translator.

Some work in this area is ongoing at CMU and elsewhere, but most
of what has been produced misses the mark.  I'm not concerned
with voice command applications.  Visualize a system that would
permit one to speak into an ordinary voice telephone, and have
a computer translate text on a screen.  I am aware of the little
constrained vocabulary trainable systems that exist now.  But
they are of but trivial value to hearing impaired people.

One approach might be to digitize the voice waveform, perform
a fast fourier transform to extract components in speech range,
try to match phonemes and lookup words.  But doing this in
realtime on anything less than a Cray would be challenging.
Maybe you can think of a fresh approach and become famous!

Good luck on your project, and keep us posted on your progress.

Jack.

... Live, from Beaver County, it's . . . .

--
Uucp: ..!{decvax,oliveb}!bunker!hcap!hnews!129!26!Jack.O'keeffe
Internet: Jack.O'keeffe@f26.n129.z1.fidonet.org