/FONT#1=BASL30/FONT#2=NGR40L/FONT#3=NGR25/FONT#4=NGR20
␈↓↓␈↓α␈↓β␈↓∧␈↓α␈↓ ¬εSTANFORD UNIVERSITY
␈↓β␈↓ ¬_STANFORD, CALIFORNIA 94305
␈↓∧ARTIFICIAL INTELLIGENCE LABORATORY␈↓ 
+   Telephone:
␈↓ ↓H
␈↓ 
+415-497-3330
␈↓ ↓H
␈↓↓␈↓ εApril 23,1974

␈↓ ↓H

␈↓ ↓H

␈↓ ↓H
Dr. L.M. Branscomb
␈↓ ↓H
V.P. and Chief Scientist
␈↓ ↓H
IBM Corporation
␈↓ ↓H
Armonk, N.Y. 10504
␈↓ ↓H

␈↓ ↓H
Dear Dr Branscomb:
␈↓ ↓H

␈↓ ↓H
␈↓ ↓HWould␈α∪IBM␈α∪be␈α∪interested␈α∪be␈α∪interested␈α∪in␈α∪broadening␈α∪its␈α∩association␈α∩with␈α∩the␈α∩outside␈α∩speech
␈↓ ↓Hrecognition␈α
work␈α
by␈α
supporting␈α
some␈α
phase␈α
of␈α
our␈α
speech␈α
recognition␈α
work␈α
here␈α
at␈α
Stanford?

␈↓ ↓HI␈α
believe␈α
that␈α
this␈α
might␈α
be␈α
a␈α
good␈α
thing␈α
from␈α
IBM's␈α
point␈α
of␈α
view␈α
for␈α
the␈α
following␈α
reasons:

␈↓ ↓H1)␈α∩It␈α∩would␈α∩bring␈α∩a␈α∩different␈α∩perspective␈α∩to␈α∩bear␈α∩on␈α∩your␈α∩problems␈α∩with␈α∩the␈α∩enthusiasm␈α∩and
␈↓ ↓Horiginality␈α∃that␈α∃fresh␈α∃graduate␈α∃students␈α∃can␈α∃offer.␈α∃The␈α∀time-sharing␈α∀on-line␈α∀graphics-oriented
␈↓ ↓Hspeech-research␈α∞facilities␈α∞here␈α∞at␈α∞Stanford␈α∞are␈α∞perhaps␈α∞unexcelled,␈α
and␈α
we␈α
are,␈α
of␈α
course,␈α
tied␈α
into
␈↓ ↓Hthe␈α
ARPA␈α
net.

␈↓ ↓H2)␈α∂The␈α∂work␈α∂that␈α∂we␈α∂are␈α∞doing,␈α∞particularly␈α∞that␈α∞on␈α∞the␈α∞pitch␈α∞synchronous␈α∞extraction␈α∞of␈α∞acoustic
␈↓ ↓Hparameters,␈α⊂on␈α⊂the␈α⊂search␈α⊂for␈α⊂new␈α⊂acoustic␈α⊂correlates,␈α⊂on␈α⊂the␈α⊂development␈α⊂of␈α⊂improved␈α∂formant
␈↓ ↓Hpicking␈αtechniques␈αthat␈αcan␈αbe␈αsafely␈αused␈αin␈αtransition␈αregions,␈αand␈αfinally,␈αon␈αthe␈αdevelopment␈αof
␈↓ ↓Hadaptive␈α
learning␈α
techniques,-␈α
all␈α
of␈α
this␈α
work␈α
would␈α
seem␈α
to␈α
nicely␈α
complement␈α
your␈α
in-house␈α
work.

␈↓ ↓H3)␈α∞My␈α∞own␈α∞long␈α∞association␈α
if␈α
not␈α
active␈α
participation␈α
in␈α
speech␈α
work,␈α
first␈α
at␈α
the␈α
Bell␈α
Telephone
␈↓ ↓HLaboratories␈α∞(1928-1946)␈α
and␈α
later␈α
at␈α
IBM␈α
(1949-1966),␈α
where␈α
I␈α
was␈α
more␈α
directly␈α
involved␈α
in␈α
the
␈↓ ↓Hearly␈α∞faltering␈α∞attempts␈α∞to␈α∞get␈α∞started␈α∞in␈α∞this␈α∞work,␈α∞should␈α∞be␈α∞of␈α∞value␈α∞in␈α∞preventing␈α
the␈α
needless
␈↓ ↓Hrepetition␈α
of␈α
old␈α
work.

␈↓ ↓HPerhaps␈α
I␈α
should␈α
also␈α
add␈α
that␈α
I␈α
have␈α
been␈α
an␈αIBM␈αconsultant␈αoff␈αand␈αon␈αsince␈αmy␈αretirement␈αin
␈↓ ↓H1966␈α⊃(but␈α⊃not␈α⊃in␈α⊃speech␈α⊂recognition␈α⊂research).␈α⊂I␈α⊂would␈α⊂be␈α⊂willing␈α⊂to␈α⊂consider␈α⊂a␈α⊂renewal␈α⊂of␈α⊂this
␈↓ ↓Hrelationship␈α
if␈α
this␈α
seemed␈α
desirable␈α
but␈α
this␈α
is␈α
not␈α
my␈α
major␈α
concern.

␈↓ ↓HMy␈α⊃principal␈α⊃concern␈α⊃is␈α⊃to␈α⊃obtain␈α⊃some␈α⊂outside␈α⊂support␈α⊂for␈α⊂our␈α⊂graduate␈α⊂students.␈α⊂The␈α⊂ARPA
␈↓ ↓Hsupport␈α∂that␈α∂we␈α∂have␈α∂enjoyed␈α∂has␈α∂never␈α∂been␈α∂adequate␈α∂and␈α∂there␈α∂is␈α∂danger␈α∂of␈α∂its␈α∞complete␈α∞loss
␈↓ ↓Hbecause␈αof␈αbudget␈αcuts␈αand␈αperhaps␈αpartly␈αbecause␈αof␈αmy␈αown␈αoutspoken␈αlack␈α
of␈α
agreement␈α
with␈α
the
␈↓ ↓HARPA␈α
point␈α
of␈α
view.

␈↓ ↓HIf␈α∞you␈α∞want␈α∞any␈α∞additional␈α∞information␈α∞about␈α∞me␈α∞I␈α∞suggest␈α
that␈α
you␈α
talk␈α
with␈α
some␈α
of␈α
the␈α
senior
␈↓ ↓Hpeople␈α
in␈α
IBM␈α
Research␈α
or␈α
for␈α
details␈α
of␈α
my␈α
current␈α
work␈α
with␈α
Rex␈α
Dixon␈α
or␈α
Chuck␈α
Tappert.

␈↓ εW

␈↓ εW
Sincerely,
␈↓ εW

␈↓ εW

␈↓ εW

␈↓ εW
Arthur L. Samuel
␈↓ εW

␈↓ ↓H
ALS:pdp10
␈↓ ↓H