[comp.sources.d] Voice Recognition/Synthesis, request for information

chuck@virgil.UUCP (Chuck Cartledge) (08/15/89)

I am looking for information about voice recognition interfaces and/or
systems.  The specific application would be a non speaker dependent
environment, with a limited vocabulary.  Something that will run in an AT
clone or PS-2 would be ideal, but that is not a limiting factor.

Also, are their any voice synthesis packages that work well given an
arbitrary ASCII string??

Any information would be greatly appreciated.

-- 
Chuck Cartledge (804)498-1012 (voice-play)                 chuck@virgil.UUCP
EDO Corporation (804)424-1004 (voice-work)
Virginia Operations
814 Greenbrier Circle, Chesapeake Va.  23320

harper@artemis.uucp (David Harper) (08/15/89)

In article <2038@virgil.UUCP> chuck@virgil.UUCP (Chuck Cartledge) writes:
>
>I am looking for information about voice recognition interfaces and/or
>systems.  The specific application would be a non speaker dependent
>environment, with a limited vocabulary.  Something that will run in an AT
>clone or PS-2 would be ideal, but that is not a limiting factor.
>
>Also, are their any voice synthesis packages that work well given an
>arbitrary ASCII string??
>
>Any information would be greatly appreciated.

I just purchased the IntroVoice VI from Voice Connection (714-261-2366) after
seeing a demo of the unit.  The board will fit in an 8 bit slot of any
compatible and comes with microphone, speaker and development software for
both the recognition and synthesis.  The recognition seemed to work pretty
well despite some background noise.  Also, the software will allow you to
tweak some of the parameters to optimize for a given environment.  As I recall,
the recognition consists of a vocabulary of several hundred words; these can
be spread out over subvocabularies to help increase the recognition hit rate.
I was not too wild about the voice synthesis half; it sounded like 'robot speech'
but it was understandable.  I was told that the demo I was hearing consisted of
a simple 'text to speech' package and, for specific applications, the output
could be optimized to make it sound better.  I have not actually used the 
system myself yet - with luck my order should arrive tomorrow.  If you're
interested, I'll let you know what I think about it after a couple of weeks of
using it.


Dave Harper    -     Convex Computer Corp.            E-mail address:
3000 Waterview Pky.  Richardson, TX 75081             harper@convex.COM
(214) 497-4525 (W)   (214) 727-4206 (H)

ivan@megatest.UUCP (Ivan Batinic) (08/18/89)

	Ditto! I'd also appreciate any information on the above
topics, particularly on the availability and performance of MITALK

	Incidentally, does anyone know of MITALK hardware? (This was
mentioned as a coming attraction in the book published by MIT on
this project.

	Thanks in advance!
	
	
<<IPB>> 

(408) 437-9700 (W) x3185

(415) 949-2226 (H)