🚀 go-pugleaf

In article <1991Mar15.171949.13836@infonode.ingr.com> 
girzong@infonode.ingr.com (Gary Girzon) writes:
> I am trying to evaluate differences between two DSP families,
> TI and Analog Devices. If anyone has experience using these
> DSP chips and the tools (eg., C compilers, emulators), please
> e-mail me.

I tnink this would be a valuable posting as well!

Herb Poppe             hpoppe@ncar.ucar.edu
NCAR                      (303) 497-1296
1850 Table Mesa Dr.
Boulder, CO  80307-3000

Hello fellow-netters,


I am interested in voice recognition systems, specifically simple minded
(and hopefully inexpensive) ones which can recognize certain words when
spoken in a sentence, such as light, move, etc. This recognition system
would be used in a PC based eliza-type program which communicates with
the user using spoken language as opposed to a keyboard. The
requirements for the recognition system are simply that it be able to
figure out what SOME of the words spoken to, it's not necessary to get
all of the words.

For those of you not familiar with AI , eliza is a program, written in
LISP, which pretends to be a psycho analyist by parroting back some of
your phrases, modified slightly. An example would be "I hate my mother"
to which eliza would reply "why do you think you hate your mother". And
so on. Normally, interaction is through a keyboard. I want to see how
hard it would be to use actual speech, both for my own amusement 
and also to learn about speech recognition and synthesis (well, I think
it would be FUN!). 

I envision a system which digitizes speech from a microphone, then cuts
up a sentences into words and tries to correlate those words (using a
cross correlation) with words in a dictionary. If it cannot correlate
any words, the program will answer with some prefabricated phrase. If it
can, it will answer with a phrase using that word. This is pretty much
all I am looking for. 

I'm not sure it will work, mainly because of two problems:

 	1. Is it even possible for a SPOKEN sentence to be successfully
broken up into words by a computer analysing a digitized signal?

	2. Will the same words spoken by different people have a higher
cross correlation than different words spoken by the same person. In
other words, do you think it is possible to decipher words in speech
samples from different people? 

Thank you very much for your help.

Sharon Gadonniex

>For those of you not familiar with AI , eliza is a program, written in
>LISP, which pretends to be a psycho analyist by parroting back some of
>your phrases, modified slightly.

I thought it was originally written in BASIC or FORTRAN (although I'm
not quite sure).  From what I have heard, it was written sometime in the
1960's, and psychoanalysts were excited for a while about the possible
applications for ELIZA in repetitive therapy (or whatever the technical
term was).  :-)

-----
John B. Melby
Fujitsu Limited, Machida, Japan
melby%yk.fujitsu.co.jp@uunet

In article <MELBY.91Apr3042124@daffy.yk.Fujitsu.CO.JP> melby@daffy.yk.Fujitsu.CO.JP (John B. Melby) writes:
>>For those of you not familiar with AI , eliza is a program, written in
>>LISP, which pretends to be a psycho analyist ...
>
>I thought it was originally written in BASIC or FORTRAN ...

It was written using a list-processing package implemented in FORTRAN.
-- 
"The stories one hears about putting up | Henry Spencer @ U of Toronto Zoology
SunOS 4.1.1 are all true."  -D. Harrison|  henry@zoo.toronto.edu  utzoo!henry

In article <MELBY.91Apr3042124@daffy.yk.Fujitsu.CO.JP> melby@daffy.yk.Fujitsu.CO.JP (John B. Melby) writes:
>>For those of you not familiar with AI , eliza is a program, written in
>>LISP, which pretends to be a psycho analyist by parroting back some of
>>your phrases, modified slightly.
>
>I thought it was originally written in BASIC or FORTRAN (although I'm
>not quite sure).  From what I have heard, it was written sometime in the
>1960's, and psychoanalysts were excited for a while about the possible
>applications for ELIZA in repetitive therapy (or whatever the technical
>term was).  :-)

If anyone cares, (can you tell whether I do?) according to (yecch!) The
Handbook of Artificial Intelligence, Vol I (Barr & Feigenbaum, '81),
ELIZA was written by Joseph Weizenbaum at MIT in '66 (Comm. ACM, V9,
pp36-45). It was written in the SLIP (Symmetric List Processor) language,
also developed by Weizenbaum in '63 (Comm. ACM V6 pp524-544). Sounds like a
LISP derivative to me, and the few snippets of rules/whatnot in B & F's
2-page description seem to confirm my suspicions.


>John B. Melby

-- 
*Standard Disclaimers Apply*|        ---Get Out Of HELL Free!---
John R. Murray              |The bearer of this card is entitled to forgive
murray@vsjrm.scri.fsu.edu   |Himself of all Sins, Errors and Transgressions.
Supercomputer Research Inst.|                                -- D. Owen Rowley