[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Head movement and ASA (and spinning)
Although one can make a plausible argument for movement as part of perception, that doesn't really begin to approach how. I would argue that an intermediate, representational stage that supports non-body-centred cognitive mapping is necessary to postulate that successive - and differing- sound fields can be integrated into a holistic grasp of a sound environment.
This was at the heart of why I was asking for opinions on sound fields and sound environments a few years ago - in the archive, I expect.. yes, http://www.auditory.org/mhonarc/2000/msg00505.html . The point was that a given environment contains uncountable different sound fields - each of which may (perhaps) be precisely measurable, but there's no real basis for assuming that this is how peception proceeds; if we were less than accurate at each stage, the accumulated error would be enormous. Therefore, it would be interesting to assume that we perform some 'quick and dirty' representation of an externalised world, and use movement to flesh out as we go along.
I remember that Barbara Shinn-Cunningham et al noticed a learning effect in reverberant rooms, where spatial detection continued to improve over some hours' exposure to a room (even for changing listening positions) and the improvement lasted at least overnight.
The question then is what is being mapped, in this scheme of things? the background-foreground distinction (crudely: sound of place, sound of things in the place) is a reasonable place to start. Again, if you make both behave exactly the same way, so as to simulate spinning of the perceiver, I think the perceiver will be disorientated.
Incidentally, a ballet dancer doesn't spin the head smoothly - usually orientating to an external feature for a limited period, then whipping the head to a new, 'fixed' (with respect to external features) position, and so on. And probably not doing much by way of auditory spatial perception, either.
Also, it is quite feasible to spin a whole sound field round, using 1st or 2nd order ambisonics. However, as the speed of rotation goes up the impression gained soon turns into that "Leslie speaker modulation" John describes. Angelo Ferrina noticed this a few years ago, that the spatial modulation soon becomes a timbral modulation. So auditory spatial perception probably becomes indistinct long before Doppler becomes distinct, in the spinning paradigm.
>>> John Neuhoff <jneuhoff@xxxxxxxxxxx> 19/12/2005 14:35 >>>
Check out Bill Chapin's company (Ausim Inc.) for some incredibly compelling
simulations of moving sounds. The strikingly realistic auditory motion
simulations that we recently used included absolute delay (to account for
the speed of sound and the changing distance between the listener and the
source), Doppler shift, atmospheric filtering, gain attenuation due to
atmospheric spreading, ground reflection attenuation, and dynamic HRTFs
from the KEMAR data set We've just completed some work on the perception of
looming and receding auditory motion that they simulated for us.
Dear Jim & Al,
The physics (and perception) of the Doppler effect depend on the velocity of
the source and the distance from the listener to the patch of the source. A
source travelling at a high velocity, but on a collision course with the
listener will not have a perceptible Doppler effect. The observed frequency
will be higher than the emitted frequency, but there will be no frequency
change on approach. So, if the listener doesn't know the source frequency,
the Doppler shift will go undetected (except for the influence of the
intensity change, see Neuhoff & McBeath, 2002).
The case of an ice skater (or a slower rotating head) is interesting. As
one ear moves toward the source (causing an increase in frequency) the other
moves away (causing a corresponding decrease). But, the pattern is reversed
after 1/2 rotation, causing a modulation effect. This is much like the old
rotating Leslie speakers that came with Hammond organs, only a rotating head
instead of loudspeaker. Regular head turns typically do not produce
perceptible Doppler shifts, probably for two reasons. First, the velocity
of the head turn is pretty slow relative to the speed of sound. Second, as
mentioned above, frequency rises in one ear and falls in the other. If you
simulated high-velocity head turns, or a source rotating quickly around the
head, this actually creates some interesting situations for ASA. The pitch
of a single source sound going in opposite directions based on the direction
of the rotation. I wonder if you could manipulate the stimulus in such a way
that the differential pitch cues would override the spatial unity of the
source and get one source to sound like two based on Doppler cues.
"...head movements for the improvement of SSL are quite a nice example of
the role of action in perception, up to the point where some say "Perception
is a behavior, a specific kind of action aiming at the driving home of a
maximum amount of information on the object of interest." (Is this
Yes, this is quite ecological and quite Gibsonian. Although "Neo-Gibsonian"
might be a better term. It's doubtful that Gibson would be much enamoured
with any of the top-down cognitive aspects of ASA, but head movements,
absolutely. It is striking to me that so much work has been done on
localization and the role of spatial separation in ASA while restraining
head movements. We've evolved in an environment in which head movements
toward a source are adaptive. If we examine the head movements of newborns
toward a source (or hemifield) one could even make the argument that they
are somewhat innate. This leaves little doubt in my opinion of their
John G. Neuhoff
Department of Psychology
The College of Wooster
Wooster, OH 44691
From: AUDITORY Research in Auditory Perception
[mailto:AUDITORY@xxxxxxxxxxxxxxx] On Behalf Of Pierre Divenyi
Sent: Sunday, December 18, 2005 7:48 PM
Subject: Re: Head movement and ASA
The major problem with the complex (=multi-source) moving source experiments
you are talking about is procedural: as far as I know, there is no algorithm
that would generate good dynamically changing HRTF's and generating even a
single moving source in free field requires quite an undertaking.
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email