You add some attractive academic thought to this problem - more
organised than my original poke.
Can I throw in another silly thought? The "training" to cope with a
modified HRTF - say, putting on a tilted wide-brimmed hat and pulling a
thick scarf round one's neck - seems to take place almost instantly. As
someone mentioned on this list before, this is probably because there
are visual clues that allow us to re-calibrate our direction sensing,
most particularly if the changes are within a range that we have often
met before. That familiarity seems necessary, because I've noticed that
if one of my ears is temporarily blocked for some reason, I can still
make the directional re-calibration but it definitely takes longer -
long enough for me to be conscious of doing it.
The silly thought is, do we just need a short-term feedback correction?
A brief visual cue, which can subsequently be dropped, because our
neural correction system retains the re-calibration until something else
occurs to convince our brain that it needs to correct again. No idea how
you might experiment with that....
Chris Woolf
On 15/02/2023 13:43, Sampo Syreeni wrote:
On 2022-12-31, Chris Woolf wrote:
It has always struck me that we can indeed adapt remarkably quickly
to local changes in our personal HTRF, and that therefore this needs
to be considered as a dynamic affair, rather than a purely static one.
By the way, there are even more remarkable examples of that
adaptability in psychophysics. Perhaps the most dramatic I know of is
the one of inverting goggles. Apparently, if you consistently wear a
headset which flips your vision upside down, in about two to three
weeks your circuits adjust to compensate, and then back again once you
stop the experiment. That happens even if you're an adult, so that
this is not an example of early childhood, low level plasticity and
the irreversibility that comes with it. (Pace kittens only shown
vertical stripes and that sort of thing.)
So how much precision is really needed for an HRTF? And how
inaccurate can it be for our normal correction ability to deal with it?
Perhaps even more to the point, what precisely are the mechanisms
which enable us to compensate like that? Because if we really
understood what they are, maybe we could take conscious advantage of
them, to rapidly train people to work with a generalized HRTF set,
instead of going the hard way of measuring or modelling individualized
head, torso and pinna responses.
One obvious answer is feedback. I'd argue the main reason head
tracking works so well is that we're tuned to correlate how we move
with the sensory input provoked by the movement. That's for instance
how children appear to learn first occlusion and then by extension
object constancy. In audition, I've had the pleasure of trying out a
research system in which different kinds of head tracked binaural
auralization methods were available for side by side comparison. The
system worked surprisingly well even with no HRTF's applied, but just
amplitude and delay variation against an idealized pair of point omni
receivers. I also adapted to it *really* fast, like in ten minutes or so.
But is there more? Head tracking, especially in a directionally solid
and low latency form, isn't exactly an over the counter solution yet.
So could you perhaps at least partially substitute the learning from
feedback with something like synchronized visual or tactile cues, in a
training session? Because if you could, you'd suddenly gain a lower
cost yet at least somewhat effective version of binaural rendering;
there would be money to be made.
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.