UNCLASSIFIED

AD NUMBER

AD002595

NEW LIMITATION CHANGE

TOApproved for public release, distributionunlimited

FROMDistribution authorized to U.S. Gov't.agencies and their contractors;Administrative/Operational use; 5 Jan1953. Other requests shall be referred toNaval School of Aviation Medicine,Pensacola, FL.

AUTHORITY

ONR ltr, 26 Oct 1977

THIS PAGE IS UNCLASSIFIED

Reproduced 6by

l'; 1" 1vices Technical Info aion agencyUm T SERVICE CENTER

4 lk Ow 7' UILOING, DAYTON, 2, OHIO

AD

11i1P' I I

/-0 11WMi

e ~SIF E~__

I* i

I

U. S. NAVAL AIR STATION

PENSACOLA, FLORIDA

RESEARCH REPORT

LOCALIZATION At:CUýACY RESULTING FROM

____--ISOLATED BINAURAL STIMULATION

1,RO,1ECT No. NM 001 )6-1.01.15

Ixv 11P (ANff y I PA 1,r,'yrr vvyI/

~~l rrle-to Vo'

0 2Cý4 CU C30

CU -0 JI

-~~~1 Q, (I"U tLO 7ý 4 ::SK .4 C . C 0 «

1 z U) lo- 00

t- wbflW 0 W~~

U)L b!U C

In U1 - . Z4 > w- OC u ,g

z 00C g U

c-0 0 d M~ C) I' ',V

Z 6 co: CoU ýb>%swlCs.ýdWE4 '0cl -0 --n4n

C) Q S d" U- gWý'U)- . 08 U)o ;,to~w 'do)~'

Cý < O -it '0

C , 00 z~L U)4~~t :3-

CO~ ~~~ w ~ 0 k~U 0 ~ 4(ý 0 to 4-P ww.'

04 PP co'0 co~' toU > j4,w~u 0 0- ~ U, N ) (01 - 0'' 1-4-Cl) .4-C) U CUb- V .CJ--'

CCw -'- 14 U0

oI~ Za; IJ oo +1 -'W )0,-~~~~ r. +1C>- ~ CS

P4 COU -Q.: ýcC4 - CU00 W oýo z co bl 0 L) r cHU)C *Nav

to ~~ 0 ,144 1

0 0 u) t Q0) W U , 0C~O

-0 og a4 S %g-oCi0. 'd 0wwC' 5C (

E'r- E' -0r. ton ~ j% 0C9

~ z .4U -0Z ~ C C)~ fj E4 H 0C0~

al~ En : A. ::s w""n Cr 00

r, S. NAVAL SCHOOL OF AVIATION MEDICINENAVAL AIR STATION

PENSACOLA, FLORIDA

JOINT PROJECT REPORT NO. 15

Thee Ohio State University Research FoundationCoouanbum, Ohio, under Contract N60R 22525,

Office off leval Research Project Designation No. NR 145-993

and

U. S. Naval School of Aviation MedicineThe Buresu of Medicine and Surgery Project NM 001 064.01.15

LOCALIZATION 0U1UlAOY RESULTING FROM ISOLATED BINAURAL STIMULATION

Report pre2ared by

Gilbert C. Tolhbust

Amproved by

John I. BlackProject Director

andCaptain Ashton Graybiel, MC, USN

Director of ResearchU. S. Naval School of Aviation Medicine

~eleased by

Oaptain J&-ec Ai. hollund. MC, USNCommanding Officer

5 January 1953

Opinions or conoltiusions contained in this report are those of the authoz,They are not to lse coastrued as necesuarily reflecting the view or theendorsement of tbe Navy Department. Reference may be made to this reportin the same way xse to published articles noting author, title, source,date, project nubeer and report number.

SUKMARY

Two hundred and five observers in groups made Judgments of apparentazimath of five different types of sound stimuli, 6 five-second presenta-tions of each sound were heard over a dual-channel stereophonic system.The microphones were mounted on a turntable capable of being rotatedthrough 360 degrees and separated to simulate intra-aural distance. Thishad the effect of being able to rotate an observer 360 degrees in relationto a-single sound source, or the sound source rotating around the indi-vidual through 360 degrees. The observers made their responses on a pre-pared form indicating by a vector arrow the direction from which thesounds appeared to come as the angle of the microphone turntable wasvaried ramdomly in relgtlon to a sing!le loudspeeker. The distance al!sovaried in four steps randomly from 8 inches to * feet.

CONCLUS IONS

The analysis of the data indicate that observers attempting to local-ize sound stimuli presented by a dual-channel sound system:

1. Were unable significantly better than chance, to localize inazimuth within a full quadrant of t 45 degrees the five types of stimuliwhen all were considered as a unit. However, the observers were able tolocalize Musc within a quadrant better than chance responses.

2. Subjects were unable to localize one of the sounds used as stim-uli within a quadrant of azimuth significantly better than any of theothers.

3. By increasing the NcorrectH azimuth angle to include t 90 degreesthe observers were able to localize some of the stimulus sounds signifi-cantly different than others.,

4. Sixteen percent of all responses made were 180 degreec away fromthe stimulus sounds. This Judgement was significant well beyond the 0.1percent level of confidence and indicates that there mea be localization,that is, confusion, whether the stimulus is in one direction or directlyopposite.

1

INRODUCT ION

The problems of souud localization have received attention by manyinvestigators over a considerable period of time (_1, 3, f_ , 9). Thepresent theories seem to indicate that localiration in a complex inte-gation of intensity, temporal, and phase differences at the two ears.Additional cues are probably present resulting from visual stimuli and/orproprioceptive muscle tonus (8). Observations have also been made thatcomplex tones are more easily and reliably located than are pure tones.

Maey of the investigations using electronic devices have employeddichotic and binaural stimulation varying one or more of the above-mentioned "fi~udamentall factors of the elements of sound hypothesized^@ neaessamy for localization. Stevens and Re== havc chown that local-ization error increases with the frequency of pure tone up to 30 ops,then decreases to a low level at 10,000 cps. They also show there is aconfusion between front and rear, particularly for frequencies below3000 cps (7). They used the technique of outdoor free-field conditionsswinging the sound source around the observer at a constant radius.

Zarlier, Steinberg and Snow had experimented with the $stereophonic"effect, or auditory perspective, using two or more microphones and loud-

speakers with the sound source (speech) behind an auditorium curtain andthe observers in the house. They found reasonably good correspondencebetween the caller's actual position and his apparent position on thestage (4). The auditory effects of a two-channel system, having twomicrophones mounted on a duwmy hea in the position of the ears and thechannels isolated to two earphones are strikingly realistic (1).

The present investigation is due to the influence of an experiencewith a high-fidelity two-channel system. The apparent positiveness withwhich a casual observer seemed to locate sound sources within a room plusthe dramatic semblance of movement proposed the question of how accuratewould the localization of various sounds be with the cues of vision, bodymovement and muscle tonus eliminated,

A statement of the problem would seem to be, to investigate the accu-racy of binaural localization using five different types of sound stimulipresented to the ears by an isolated dual-channel system matched through-out. Under tho conditions of the present experimental techniques theworking hypotheses are:

1. Observers will not be able to indicate azimath of a sound sourcewithin t 45 degrees at better than chance frequency.

2. There is no difference .among the five sounds used as stimuliwhen the criterion measures are the frequency with which observersindicate azimuth within't 45 degrees, (also within t 90 degrees).

3. There is no direct reversal (180 degree displacement) ofobserver' s judgments,

2

E0J 'PM•T

Figure 1 is a block diagram of the dual-channel sound system, eachchannel isolated, with the output going to each earphone of matched head-sets at the observer' s stations located in a soundproofed room. The mini-mum ambient noise level in the room was 46 db (C scale on the GeneralRadio sound level meter).

The stimulus sounds were recorded on a magnetic tape recorder (Ampex400) and were played back upon the same machine. This recorder was locatw-in a control room and fed a 12-inch loudspeaker (Western Electric 728-B inan enclosed 3 cubic foot baffle) located in a small sound-proofed room.The loudspeaker was mounted so that the center of the speaker cone was atthe game height above the hortutal a* the d al ierophones end deliveredan average signal of 78 db, rc .0002 dynes/ca'S, at 8 inches. The twocondenser microphones (Altec 21-B) were mounted on a brass turntable 5inches apart, simulating inter-aural distance. The turntable rode twotraverse rods parallel to the h'rizontal. This allowed the microphones tobe rotated through 360 degrees &ld to move in a straight line from theaxis of the speaker cone a distance of 8 inches minimum to * feet maxi-mum. The loudspeaker-microphones relationships are shown in Figure 2.The microphones led to two matched pre-amplifiers in the console (Altec250-A), located in the control com, which passed the signals to two"identical" amplifiers (Stancil-Noffman R-48-P). Each amplifier droveone of the earphones (ANB-H-1) in each of the fifteen headsets located inthe larger sound room at the observer' s stations. The signal under theearphone cushions as measured by a probe tube coupled to a calibratedcondenser microphone was 80 db when the microphones were 8 inches from theloudspeaker.

In essence, the signals were presented by a single sound sourcepicked up by two microphones, each amplified by an identical amount andpresented to separate earphones. The system was so arranged that when thepointer on the turntable was aimed along the axis of the loudspeaker,pointing toward the loudspeaker, the microphone on the right, or 90 de.grees, fed the channel leadJl-to the right ear of the observer. Themicrophone on the left, or 270 degrees, fed the left earphone. The S/Irati- ,-f the total system was 21 db at the stimulus level used.

PROCEDUME

Two hundred and five young adults acted as observers in groups of 4to 15 individuals per group, listening in quiet. Each observer madeJudgments concerning five different types of sound stimuli and each typeof sound was presented six times for five seconds duration each presenta-tion. Between each five second stimulus there was a five second silentperiod. Each 6aries of like sound stimuli (one line on the responsesheet) was announced, telling the type of sound stimulus to follow.

The microphoneo wiore oitchod off at tho control room after the

initial announcement of each line and betweeL each five second stimulus.This allowed an experimenter in the smaller sound room to rotate the mic-rophone turntable to a designated azimuth angle and to vary the distancein relation to the loudspeaker to one of four positions, i.e. 8 inches,1 foot, 2 "nd * feet. The azimuth and distance varied in a random orderfor each type of sound. The schedule is found in Appendix X. The fivetypes of sounds used as stimuli were: (1) 500 cps pure tone; (2) sus-tained vowel Ca] 125 cps sung by a male, baritone voice; (3) continuousspeech, a paragraph of factual prose; (4) musical passage of full orches-tra, symphonic arrangement of "Little Brown Jug"; (5) white noise.

Each group of observers received instructions to assure that theproper earphone went to the "correct" car. Each individual was checkedfor earphone orientation. The directions for the use of the responseform, (see Apuendix IT), were informally presented but always containedessentially the following informationg

"You are going to hear 5 different types of sounds.Each of the five sounds you will hear six times and eachwill last 5 seconds each presentation. Notice on yourresponse sheet (see Appendix II) that there is a repre-sentation of a head for every time you hear a sourd.These 'headst are supposed to ripresent your head,looking down from on top and facing in the same directionyou are facing. The bump toward the top of the page youare to imagine as your nose. The two bumps on eitherside of the 'heads' are supposed to be your right andleft ears respectively. The dot in the center repre-sents the top center of your head. The sounds you willhear (these were enumerated) may appear to be comingfrom various directions, in front, behind or from thesides. Will you indicate by a line, or arrow, the dire-ction from the center of your head, i.e. the dot, fromwhich the sound appears to come. If the sound seemssome distance from you make your arrow longer than ifthe sound seems closer. If the sound appears to stayinside your head and you cannot assign a direction toit, draw -a circle around the dot. For each line ofheads the recorded voice will tell you the type ofsound you are to listen for. Use a new 'head ' foryour direction judgment each time you hear a fivesecond sound".

Several illustrations of the use of the vector arrow to reportazimuth judgments were given to each group, Preliminary experimentationhad indicated that it was difficult for "naive" subjects to make bothdirection and distance judgments so that the observers were told to ig-nore attempting to fill in the 'feet and inches" blanks on the responseform and to respond to distance changes if they could, using the roughindication outlined in the "Hirectionis" above.

4

DATA

Four tabulations were made from the responses. These are found inTable 1 showing the frequency as well as the mean frequencies with whichthe judgmente of tho 205 observers, each making 30 responses, fit one ofthe four categories that were employed in testing the hypotheses.

Chi.9g.._. tests were run to determine if the frequencies with whichthe observers made "correct" responses within k 45 degrees were made betterthan chance. The individual chi .. g_ e tests made for each of the fivetypes of sound stimuli showed that Music had a value of 5.77 which issignificant at the 2 percent level. This was the only significant value;the next largest chi square fell somewhere between the P0 - 30 percentlevel.

Two analyses of variance were formulated from the data. The firstmatrix used as basic scores the frequencies that each individual was ableto indicate "correct" azimuth with i 45 degrees for each of the five typesof sound stimuli. The second analysis used as basic scores the frequencythat each response judgment fell within k 90 degrees. Both analyses aresummarizod in Table 2. Since each matrix had the same degrees of free-dom that column is common for both.

Table 2

Source of Sum of Sum oflice Suanres Squares &f. Variance Variance () (F_)

+450 + 90 ± 45 +- 900 t 450 "900

Betweenconditions 11.18 23.54 4 2.80 5.86 1.98 5.44 *

Between

subjects 776.42 998.46 204 3.81 4.89 1.92 * 4.79 *

Residual 5 799.60 816 1.46 1.02

Totals 1i%45.66 1821.60 1024

*(Significant at the 1 percent level of confidence 1.28, 204 and 816 dofe)

P**(Significaut at the 1 percent level of confidence 3.35, 4 and 816 d.f.)

The value for (j) at 4 and 816 degrees of freedom would have to begreater than 2.58 to be significant at the 5 percent level of confidence.It would seem, therefore, that using the present experimental conditionsand the criterion of * 45 degrees that the (I) ratio of 1.98 indicates nodifferences due to the types of sounds used as stimuli. Subjects differedsignificantly, (F) of 1.92, in their ability to localize sound within anarc of 90 degrees.

5

Tne second analysis indicated that as the "correct" azimuth anglewas increased from ± 45 degrees to a t 90 degrees that the types of soundsused as stimuli differed significantly as to the frequency with which theywere localized. The (L) ratio between sound conditions of 5.44, signifi-cant at the I percent level of confidence, makes untenable the hypothesisthat sound types do not differ when observers attempt to localize themwithin a 180 degree arc. The (F) ratio indicating subject variance wasalso highly significant for the ± 90 degree criterion.

Because of the significant (_) found in the second analysis of vari-ance aboves (t) tests for related measures between the various types ofstimulus sounds were made. The basic measures were the frequency withwhich the responses fell within ± 90 degrees. Reference to Table 3,showing the (t_) ratios between the responses to the five types of stimu-lus sounds, indicates highly significant differences between music andContinuous Sceech (t of 4.538). There were significant differences be-tween Music and EOQ (Q of 2. 869), Music and o22nq (I of 2.794) andContinuous Sceech and White Noise (I of 3.192). The data do not showstatistical differences between Music and White Noise or Tone and Cal

By inference, then, it would seem that the more complex sounds, i.e.

Music and White Noise, are judged equally well and the simpler sounds of500 cps Tone and _.j also judged equally well are different as to"accuracy" of localization within 180 degrees. If the evidence of thechi sguare test showing Music to be localized better than chance and theexamination of the mean responses is admitted, it is possible to extra-polate that Music may be on one end of a continuum and Tone or [a]near the other.

Most of the investigators reporting previously on localization haveindicated the difficulty experienced by observers in "hearing" the soundsdirectly in front of them or directly behind. Often they would judge thesound to be 180 degrees away from the stimulus. The frequency with whichsuch direct reversals occurred in the present investigation were tabulated.Sixteen percent of all of the possible judgments fell into this category.

An examination of the response forms during tabulation indicated thatthere was no trend for one individual responding to the five types ofsound stimuli to consistently make a direct reversal in localization.They appeared to be randomly distributed and not confined to the conditionin which the sound stimulus came from directly in front or directly be-hind. A chi aguarotoat to determine the ratio indicating responses sig-nificantly better than chance yielded an X t of 66.3, well beyond the 0.1percent level of confidence.

t- (980-768. 7 5 )2 (5381.25-5170)2= , + -- 66.3768.75 5381.25

These results would seem to uphold the observation made by previous

6

investigations that direct reversel. judgments occur frequently.

An informal observation was made during the time of processing thedata. It was noted that when the stimuli were less intense, i.e. 2 and* feet from the loudspeaker, that the number of responses indicating nolocalization of the sound but centered within the head increased. Onlyinfrequently did the individual make such a response when the microphoneswere at 8 inches or at 1 foot. The few "sophisticated" observers thatlistened prior to the more formal investigation had made opposite reports.

DISCUSSION

The results of the present experiment indicate that eliminating mostof the cues to localization other than directing sounds to essentiallythe tympanic membrane widen the localization accuracy of azimuth anglepreviously reported. It seems that the next logical step would be tointroduce some acoustical impedance, i.e. a shape representing a head,into the present system. It may be that some shape such as the head could,by modifying the phase and intensity relationships of the present equip-ment, provide additional cues for more accurate responses.

The present experimental setup should also provide a reasonableapproach to a study of distance localization. The observers seemed tofind that making two judgments on the response sheet used in the timeinterval of 5 seconds were all but impossible. A separate experiment inwhich distance is the sole response seems feasible, both with and withouta simulated head interposed between the microphones of the two-channelsystem.

An additional group of studies that seem plausible using the high-fidelity stereophonic equipment would be amount, or degree, of movementof a sound stimulus in motion. The studies could be done with or withoutusing a reference (stable angle) sound. Prior to any investigation re-garding movement (apparent or actual) perhaps azimuth angle localizationstudies could be redone using a constant reference sound. These are afew of the studies that are possible using the present, or slightly modi-fied equipment.

One of the circuit refinements that needs to be incorporated into thepresent system is a gate circuit to eliminate clicks that occurred infre-quently in the present instrumentation. These extraneous noise "clicks',may have aided or hindered localization of the particular stimulus sound.The noises, however, occurred randomly and infrequently.

The results of the present experiment tend to support the theorythat a more complex sound such as music stimulus can be localized moreaccurately than less complex sounds. The lack of a statistical differencebetween Music and White Noise lends further support. Contra-indications,however, are the wide differences in the (t) ratios between Tone.- Iusic,

7

Dj_ No1se - L4L.e and Mus -i hXIt ULo_. The trend is towazd more sig-nificant differences between a complex sound and a less complex, than be-tween "simple" sounds or complex pairings.

8t

BIBLIOGRAPHY

1. de Boer, K., Stereophonic- sound reproduction, hjlis Technical Review,1940 V, 107.

2. Boring, E. G.,9 Histor qf~ F Pycholoa, X. Y., D.Appleton-Century, 1929, 108, 288-309, 482.

3. Miller, G. A., Rosenblith, W. A., Galambos, R. and Hirsh, I. J.,A Bibliograiohy in Audition, Cambridge, Mass., Harvard UniversityPress, 1950, Vol. I, 1950 Classification, 19.

4. Steinberg, J. C. and Snow, W. B., Physical factors in auditoryperpebctive, E System Technical juaol, 1934, XIII, 247-260.

5. Stevens, S. S., Experimental Psychology, N. Y., John Wiley & Sons,1951, 1026-1030.

6. Stevens, S. S. and Davis, H., gEaring, N. Y., John Wiley & Sons,1938, Chap. VI, 167-183.

7. Stevens, S. S. and Newman, E. B., The localization of actual sourcesof sound, American Journal of Psycholog, 1936 a, XXXXVIII, 297-306.

8. Wallach, H., The role of head movements and vestibular and visualcues in sound localization, Journal of Experimental Psychology, 1940,XXVII, 339-368.

9. Woodworth, R. S., Experimental Psychology, N. Y., Henry Holt, 1938,518-538.

0~02

4$ 0

00

o43

to- r- s0t

.40) it)

0

a)

r4 0 a) 0

U- C) a)I to00 -

0co

43 0Jn to r-4

0 H(() V-4 CDC'QC- 4

0 01

43 0

00

o b4 0 l CA ) C\) Q

4) "94

43 CA4 0)-4 CA

4-')1 1-14 U)C' ?oCl40 0) '.')0 r-4

Id 431 LO0 LO ) -

0~

0)

W04 400F

Q ) 0 0)43 0)l +1 43

u-0 4-4 W 4l -Hr 0 0W

to 0 -r4 w. WJ.0) A) 0 0A E

r-iO 4-4 P11P1 P4

Table 3. (t) ratios for related measures and the level of confidence forrejecting the null hypothesis between each of the five sound stimulusconditions when the basic measures were the frequency of the responsesbeing "correctu t 90 degrees.

Stimulus SOu ( Confidence Level

music and continuous speech 4.538 0.1 percent

white noise and continuous speech 3.192 1o0 e

music and U 2.869 1.0 N

music and tone 2.,794 1.0 N

white noise and tone 1.739 10.0 N

continuous speech and •L-) 1.6870 10.0 '.

continuous speech and tone 1.639 20.0 N

white noise and a3 1.143 30.0 N

tone and [d] .070 above 30.0 3

music and white noise .038 above 30.0 *

-, I-

oo

E

4z-

0 C-.

C3

- *L

P4O'17

14e F1~

w03*

oa) 014

4D4

... ... .......... ... .... 4 IX1.. ~ ~~~ ...... . .0~~. . . . . . ...... .. ....... .....

............... . . . . . . . .. . . .

ii, -~ .. *............ P.Ik XJ...... ~\\ ;) ? .. ... . . . .. .* ***................. . . ... .. . ..

... ~ v* 4.. . . . . . . . 4 * *

. 44.e...... ; .. b. t ta t,. .s. &

I

*APPEDIDX I

A4ZIMMFH ARJD DISTANCE SCHEDML

1. Condition I - 500 cpu tone 4. Condition IV - Orchestral music

0o at am 2700 at 31

2250 at *1 1350 at 21

900 at 2' 2700 at I1

1350 at 11 00 at 8W

-450 at 80 3150 at 8"

3150 at 21 2250 at 21

2. Condition II - vowel Ed-1 , 5. Conition V - Tite noise

450 at 3JI 900 at 1i

2700 at 8N 450 at 8"

1800 at 1' 1800 at 21

0 at 8" 1350 at, 2'

3150 at 21 2700 at 8"

900 at 2• 2250 at 3*

3. Condition III - Continuous speech

225 at 8"

g9o at 3if

00 at 2'

315° at 2'

1800 at 80

45 at 1'

A"IMEIIX 11

Job No. 113

BIKNURAL SOUND LOCALIZATION

NAME_________ Agow___I

Have you aver .been -xposqd to prolonged gunfire ?Have you avar b-en exposed to prolonged aircraft noise_ 3 As pilot?Q-oý crow or flight deck_.?

O0 00.)

0Q () ( 00 -.t_, _tzf_~z r _tnf_~

Itii I _tJ ~f ~_ft__m I&

-Jt-ýA-it-I

Reproduced by

irmed Services Technical Inform-ation 9g e pHyDOCUMENT SERVICE CE .,NTER

KNOTT BUILDING, DAYTON, 2, OP,'t it

Uik I A Ii' F1 E