CONTRACT RD-94 TASK ORDER NO. 2

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78-03300A001900110015-2
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 4
• Document Creation Date: December 22, 2016
• Document Release Date: May 3, 2011
• Sequence Number: 15
• Publication Date: May 15, 1957
• Content Type: LETTER

X" 15. 957
Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-0330OA001900110015-2
sab j.stts P ra t
Task Oro. N. 2
25X1
Ia aseardatao with Avtlsle 2 or thr basle oestrast,
th. a are forwarisi herewith two (2) eopl.s of the MonUft
Proge os t for ap~rll, 1q5 an ask tea' no. a or .
The rop"t to 4 te& April, 1+ 7. "Nor% to ~IflID.
An tlasal a~T is being hell by the we
..gisr for UO ass at w P" ms 40 ahf at this lalins.
to .a sotlou with this s thlp prasross report, the
rollewi*g at'oruatiau is oub"tt"s
Total expaimUtume to 3-31-57
+I ate edsslt nt? as of
T.sida romalulm as at 3-31-57
Asoistant
sIMIM
ftoloo roe
$ev. +t contra" AdMINAStratiou
Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-0330OA001900110015-2
 Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-03300A001900110015-2
CONFIDENTIAL
1bnthly Progress Report
April 1957
Task Order No. 2
Contract No. RD-94
Audio Noise Reduction Circuits
The object of this project is to develop a noise reduction
circuit suitable for use in separating speech intelligence from a signal
containing speech and noise when the speech intelligence is masked by
the noise. The proposed method involves a principle which has been used
successfully to improve the signal-to-noise ratio in music reproducing or
transmission systems.1 The system used for music contains bandpass filters
which pass frequencies over a range of an octave or less. These filters
are used at the input and output of a non-linear element. The output of
the non-linear elements contain the fundamental, and also harraonios and
subharmonios of the funda.ental. However, since the pass band of the
input and output bandpass filters is no greater than an octave, the
harmonics and subharmonics are not transmitted by the system. The
funotion of the non-linear element is to reject all noise signals below
a given amplitude or threshold love!. The threshold levels of the non-
linear devices in each channel can be adjustors so that, in the absence
of desired signal, the noise is rejected. ion the desired signal is
greater than the threshold level, the nosy-linear olemonts allow the
composite signal to pass. Thus, for passages of recorded ?Tunic, when
the music signal is below the noise level in a given frequency channels
the channel is inoperative, and its output is eliminated from the total
output. Since the contribution of this channel to the ~:otal output would
have been only nois-,, the overall noise level is reduced. When the music
1. H.F. Olson, t?Electronics," Dec. 1947.
"f lNFIDEI\I\L
Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-03300A001900110015-2
 Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-0330OA001900110015-2
CONFIDEN -~
signal in a given c;3-:eilnol is greater th!n the noiso, the channel eonchiots
and allow the composite signal to pass. Thus, a channel conduots only
when the desired signal. is greater than tl,e noise, and rejects vahen not o
alcno is present.
In order to apply this me thcd of noise roduotion to speech,
when the wide band. spec c1: ratio is y::ry low, it is
ne scary to find fi'o'qu3Tncy 2'C:'" ions i , which t.he)re are tim'e's v hon the
speech anplitudo is greater than the noise. Although the long t1-':.c
r.ve 'are -pec-'r of spec ch. is and u::~rt .l^.r in chafe to u:.a
speetlnkr. Of room T'oise,z U-1-o ~?hort tine spectrum of various speech sounds
contains re ions of xaximum a orgy Cal ed speech fori:'ants.3 The aSal-LLT. 'ion
that e athcd of noi.3G could be util:i?od for speech as based
. si.,to to find freyc.3ncy a?og cps 'c1
u-..'.on the be r ?Sf t! at it : oul d b cs
s
1.1, .. ,: A' :pitv+( ?. of e fGr : LIi'GS 1 1 ` ` e3I' than the
??ltO.tl {Ln: tFi Y1 C.J~+1 C'1 i^~Ld..C_ r t ..la-' 10 ..30
a > Ls 1 ant . ~.t par?, of the Vice.
14. stv.dy has been ait~ d '60 determine wha ., I lYiCiwid t ;#3 are r'ciju .red
crc or to obtain sr J.;eii 1`037 .._._ t .c.rpla.. n'eo ab:ve 'c':.E; noise lion a
la-rd speech ca.ipl-: l:'.(; ; ills .:~__i1:1b]._ in ? t; i3 :kacv,11 that, for
noise 1'i?_t'i: E) CCS?1;?~ oT'r~JlT v.t:l 'i t ?_3 the cc-,': ~)onentrj in the
iI:nod?ate 2.-zquoney of .'.''_:'t : aoP ::t;1 tone ."I ti.ch contribute to the
Lking.4 Y7a73n_ a ver-i-' nex':t`c)- Land of :eoiso is iced to mask a pe.'1''e toms,
the Titr.X2{2.?le-, i c ..aces' as the ::.?;;(: i?Ridk; '.. Until a coYtf:l.:i.il baalC:':iild~i;il
IS reached. :I-?te thin, a3 tic l ba .Cwi: th is i.ri.c;rL:-2 t: '.'70 amount Of
rr,C:a.n rt::w:z:?,a c,.,~~>~:.~ntn e.~.v ..e.., : ~.,_ ~_~ ?t .r .,...rstl .,..,. .=c:u?.. r rdcc. t33 rti
fi_-Z:d++1g':...>:c Measurements have been
a,..iCj. On `i,; .} t! a 'SU: ti tt~
1a. l~s 1(.tG_:(3$'~ 'r:T~,vTvj"! .'?.t I..}~>v 'a J;-`.'.. :CS:'.v.. t(.-if ~ T.c.t^ ,c
I:c?' York, 1956 (soc Zt i` u:res T')! cud 70).
3. Op.cit. chap. I.
4. L.L. Beranek, "'Tie l)osign of Speech Systems," ;Uroc. :QT'
Vol.. 35, pp. 882, Sept. 5. N.R. French and J.C. Stoinwergr ".Factors (=overning; the Inttolligibil.ity of
Spocch Sounds," Jour. Acoust.BcC `Tol. 19, Jan. 1747 see Fig. ro `j)
' e N A 5 t
Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-0330OA001900110015-2
 Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-0330OA001900110015-2
CONFIDENTIAL
made using filters which were both narrower and wider than the critical
bandwidth. Both pure tones and speech mixed with continuous spectrum type
noises have been studied, The results of this study show that, for the
narrowest permissible bands which can be used to pass speech formants, the
number of times the speech fornsnt amplitude in a given band exceeds the noise
is small. Also, in these bands, the speech amplitude is never considerably
greater than the noise. Since very narrow bandwidths are required to reduce
the noise below the signal, the number of bands required to cover the speech
spectrum is quite large. There is no satisfactory way of evaluating the effect
upon speech intelligence of small contributions from many narrow bands without
building a nany channeled circuit and evaluating it. From the information
available from studying a few channels throughout the speech spectrum, it
seems possiblo that some improvement in intelligibility can be effected, but
this improvement may prove to be small.
In view of the fact that there is no convenient way to evaluate the
contributions of a few narrow band channels to speech intelligibility, a complete
nulti-channel system is being developed in order to determine the effectiveness
of this method of improving speech intelligibility in noise.
During April the wiring and model shop work on the eighty channel
noise reducer was completed. The chassis have all been checked and perform
satisfactory. The individual channels are now, being alignedo The frequency of
each channel is adjusted to within + 2 cycles per second of the design frequency.
The frequency is measured with an electronic counter to insure accuracy. The
bandwidth of each channel is adjusted to within -+ 2 cycles per second of the
desired bandwidth; that is, 3 db less than the critical bandwidth at each
frequency. The power supply has been installed in the rack and the associated
wiring is commpleted. As soon as all of the chassis are aligned the complete
noise reducer can be evaluated using speech.
~R:pas `~c N I D E N I A L
Sanitized Copy Approved for Release 2011/05/03: CIA-RDP78-0330OA001900110015-2