perception of direction of sounds
Didier A. Depireux
didier at src.umd.edu
Fri Sep 1 10:31:39 EST 1995
>>Well I'm really only interested in the underlying acoustic information..
>>that the brain makes use of, not in the way it processes it. Replies..
>>from others and also replies to a similar question I posted in..
>i can't picture what plane you're talking about, but am very curious. could
>you explain it, please?
What he means are not planes, but what we call the "Cones of confusion"
(which are really hyperboloids, but anyway). The question is: when you hear
a sound, how do you know if it comes from behind you, above you, under you or
in front of you?
In theory, a source located in the median plane ought to cause complete
confusion, since there are no intensity or phase differences at the
ears. However, it's been shown a long time ago (Stevens and Co, in 36!)
that confusion is pretty bad at low frequencies (under 2.5 K) but low at higher
frequencies (above 5K). Localization gets better if the source is away from the
median plane, though.
There was an poster at this year's ARO (Association for Research in
Otoloaryngology) showing that in the case of a cat's ear, if you present pure
noise to the ear, the spectrum that is transmitted to the eardrum depends sharply
on the origin of the sound. So a sound coming from behind the cat, for instance,
will have some narrow range of frequencies literally removed (say from 3200 to
3350 Hz, for instance) (removed in the sense attenuated by 20-30 dB).
Middlebrooks is really the person who would know about all this stuff.
He published an article in 87 with Knudsen in which he showed that the
directional sensitivity of cells (someplace along the auditory pathway, in the
IC, I think?) can be changed by changing the shape of the pinna (ear) closest to
the sound source.
So anyway, even though sound localization is not a well understood subject,
it is probably done through experience (i.e. it's not something that hard-coded
in the brain) and it's done by comparing the spectra received by the ears.
Notches in the spectrum received by one ear, and the absence of those notches in
the other ear, signal that the sound comes from a specific direction.
In our lab, we have a tape made by Dick Duda, where there's a recording of a
plane flying by the mikes. The sound gives the impression that the plane is
taking off right in front of you. Then he plays the sounds again, but low-passed
under 8k. The plane sounds like it's a little lower. Then he low-passes the sound
under 5K, I think, and it sounds like the plane is passing under you...
didier at src.umd.edu
More information about the Audiolog