7412533507

I

Fig. 4. The human ear is ▼ery sensltbre to phase differences below about 1500 cps. A lead or lag by 100 n*ec. of a peak at one ear usually per-mits a elear judgment of "oH mid linę” modę.

A =.0001 SECOND

tensity cues come into play, performance deteriorates so sharply that a cery inefficient minimum audible angle of 10 degrees exists at 3000*4000 cps while directionality is. for all practical purposes, complctely lost at 7000 cps and ahove.

Evidently the relations among head siatę, head shape, and wavelength are not such as to make intensity cues cery uscful in anything but the sym-metrical mid*line condition. This fact must Itc considered when placing microphones and speakers to create the stereo illusion.

FRE0UENCY

Fig. 5. The head casts a sound shadow that attenuates the higher frequencies.

Fig. 6. The stereo illusion in pure form em-ploys two microphones in a dummy head.

SOUND SOURCE

AMPLIFIER    AMPLIFIER

of the human. The lowcr froquencies tend to hond nround tho head so thal the sound ''shadow" cast by tho head introduccs a relative bass-lłoosl to the spectrum rcaching the farther ear. (Sce Fig. 5.) Also, there are reinforce-ments and interferences brought into play by the configu rat i on of the head and pinna. Notę that the sensiticity of each ear bt/ itself vaHes with the direction of the sound. The hinaural condition must com pensa tc for these differences. In animals with mocing pinnae, these variations are probably magnified quite deliberately and used to provide further information.

A quick calculation will show that the intensity dilTercnces arising front movements of the head (or. what amounts to the same thing—moving the source ofT mid linę» are much greater throwgh many octaces than the differential loudness discrimination available to the average human ear. An 800-cps tonę from one side casts an 8 db "shadow" on the oppositc ear. while errors of mid-line locali/at ion can be quite negligible on the basis of the 0.5 db discrimination i>osscsscd by many individuals.

It is necessary to emphasizc what we havc stated previously that two dis-tinct mechanisms underlie our sense of directionality—and these in differ-ent frequency regions. A granh show-ing the minimum audible anglc. in degrees, to which the average head is sensitivc in the median piane rcccals a broad minimum at about 1 degree between 250 and 1000 cps and a rapid rise abovc 1000 cps to a maximum of about 3 degrees at approximatelv 1500 cps. For these lower frequencies. tem-poral characteristics ttime of arrhal. phase l underlie performance. Ab»ve 1500 cps. the minimum audible anglc in the median piane fluctuatcs around 2 to 3 degrees up to 10.000 cps. For these higher frcqucncics the localization cuc is provided by intensity differences (the sound “shadow"i.

When the minimum audible angle is measured for sound arricing at the head at an angle 45 degrees oft the mid linę, performance is generally similar to the mid-line condition but 1 to 2 degrees worsc. up to about 1500 cps. But for higher frcquencies. where in

To procide the strongest possible stereo illusion it is only necessary to re-create in a pair of ears. by any con-venient means. the time and intensity conditions generated at a point in space by an actual sound source.

Re-creałing the Illusion

Probably the most compelling illusion of a wide-azimuth sound, or of a mocing sound source, is created by placing two omni-directional microphones in the simulated ear canals of a dummy head and recording the out-put on dual-channel tapes. This program materiał is then fed to two ear-phones. (See Fig. 6.) Here all cues hcard by the ultimate listener are as if his head were in the position occu* pied by the dummy head and the illusion is complete.

If playback is to be through loud-speakers. other considerations are in-colced. The original recording studio and the listening room must exhibit certain basie similarities. The speakers must not be too far off mid linę in relation to the listener. They must be matched closely in phase and the two channels must be symmetrical to with-in a few microseconds in time dclay.

Fortunately. these rather stringent conditions need not be met 100 per-cent in order to create a satisfactory illusion. A considerable blurring can be tolerated without destroying the illusion and. of course, the two ears will always appreciate the added "liceness" and volume procided by the second channel. But it is very easy to mistake "liceness" and colume for true stereo illusion—which once heard in all its purity is an unforgettable experience.

This blurring can be studied in the anechoic chamber. If one places two microphones four inches apart then records a wide-azimuth sound, when replayed through a two-speaker array there will be a cube (about one foot on a sideł where one can “immerse" the head and experience the stereo illusion. If the microphones are placed about a foot apart. the illusion volume will be a cube about a yard on a side. How-evcr. the illusion on the fringes of the cube loses some of its force—rather soon one encounters diminishing re-turns and the reasons are not hard to

39

October, 1959