COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BAS35.FNT[XGP,ALS]\M1BASI30\M2NGR40\M3NGR25\M4NGR20\MFSTA200\
C00010 ENDMK
C⊗;
\\M0BAS35.FNT[XGP,ALS];\M1BASI30;\M2NGR40;\M3NGR25;\M4NGR20;\MFSTA200;\;
\'3;↓↓\FFS\F2
\'3;↓Q\CSTANFORD UNIVERSITY
\F3\CSTANFORD, CALIFORNIA 94305
\F4



ARTIFICIAL INTELLIGENCE LABORATORY\←L\-R\/'7;\+R\→.\→S   Telephone:
COMPUTER SCIENCE DEPARTMENT\←S\→.415-497-4971
\F0\CJune 26, 1974

Mr. Bertrand Delgutte
89 rue du Marechal Delattre de Tassiony
59290 Wasquehal, France

Dear Mr. Delgutte:

\JProfessor McCarthy has asked me to answer your letter to him of May 21, 1974.

We have a very small research effort on speech recognition at the present
time, largely because of restrictions on funding. We are doing nothing on
synthesis altho some work is being done along these lines by another group
that has a practical application in mind. I happen to head the recognition work
and we have two post-doctrate employees who have an interest
in speech work but who are currently spending most of their time on other things.
There will be at least two graduate students from the Computer Science
Department and perhaps one other from E.E. working on speech recognition in the fall.
Activities will be at a low ebb during the summer as I am going to Europe for a
speech meeting in Stockholm and for an extended vacation.

Very little has been published by our group in the outside literature although
we have submitted several reports to the ARPA group, which will be described
below. Much of the earlier work has been superseded by more recent work by
Professor Reddy (who was in charge of this early work at Stanford) and by his
students at the Carnegie Mellon University, where the same general approach has
been continued.

The main thrust of our present work is on the acoustic end of speech as contrasted
with the emphasis of much of the current work in
the United States which has been concentrating
heavily on the syntactic and semantic aspects. We are concerned with speech
recognition relying heavily on the acoustic signal. We have been trying to 
develop tools for determining useful acoustic correlates and we have been
collecting data along these lines. The emphasis is on continuous speech and
on developing techniques that do not presuppose a very limited vocabulary. Our
facilities for this work are extremely good as compared with those at other
research centers.

There are several groups here at Stanford working on the general field of
speech. We have a group here in the Artificial Intelligence Laboratory that
is interested in the general problem of language understanding. Then there is a
group in the medical school with interests in the human aspects of speech production
and understanding. There are several people in Electrical Engineering with interests
and occasional students working on speech recognition and on speech synthesis.
There is a group in still another department with an
operating speech recognition system that is
used to teach arithmetic to small children. This is an adaptation of a system that
was developed in the Artificial Intelligence Laboratory, interestingly enough, by
a French student, Pierre Vicens, who received his PhD. from Stanford and has since
returned to France.  We have still another French student,
Bertrand Meyer, with interests in speech
here at the present time but he is leaving in a few weeks to return to France.

As you probably know, there are a number of groups at various places in the
United States supported by the government ARPA organization and these are all
interconnected by the ARPA network. We work closely with these groups
and our computer is also connected to the net. With these facilities for working
with others, the amount of collaboration that one does depends on one`s desires.
In fact the danger is that one can spend one`s entire time in keeping informed
of the work of others and so leave little time for one's own work.

I am at somewhat of a loss as to what to tell you to read, there is so much.
Perhaps if you were to get a copy of the preprints of the Speech Communication
Seminar to be held in Stockholm August 1-3 from the Speech Transmission Laboratory
of the Dept. of Speech Communication at KTH in Stockholm, this might be a
start and then by going backward through the references in these papers you
could follow your special interests back as far as you like. Most of the
recent work is only available in conference reports which are sometimes hard
to get. The Journal of the Acoustic Society is another good source.

I believe that this answers most of your questions.  I can only add that I will be
glad to talk with you on your arrival and perhaps we can work out a program that
will interest you.\.

\←L\→S\←R\-L\/'2;\→L
Sincerely yours,




Arthur L, Samuel
\←S\→L
ALS:pdp10