Re: Interaural Level/Time Differences.

["Richard F. Lyon" <lyon@xxxxxxxxx>]

Ray DeGennaro wrote:
>> I'm looking for a (ideally) a transfer function from a signal at a given
>> azimuth  to the interaural level/time(phase) difference at each ear as a
>> function of head diameter, etc.

Richard O. Duda wrote:
>Phil Brown recently completed a MS thesis for me, in which he included the
>Matlab code for a headshadow model originally developed by Dick Lyon.  His
>thesis can be obtained in pdf format via
>
>http://www.geocities.com/SiliconValley/Heights/5222/thesis.pdf

Actually, it's a modification of a headshadow model developed by Dick Duda,
but tweaked and Matlab coded by me, by fitting to KEMAR-head data recorded
by Duda.

There are available theory and programs for the IID and ITD effects of a
"spherical" head (and now corroborating experimental data on actual bowling
balls by Duda), but for real human-shaped heads it is too complex to
present as anything other than big tables of measurements--until we do
a better job of modeling it, at least.

The headshadow model in Phil's thesis is actually the one we use in Apple's
"Sound Sprocket" implementatin of 3D sound.  It is parameterized only by
interaural angle (azimuth in a system with poles at the ears) and has no
vertical (around a cone of confusion) cue at all.  It is a very simple
filter (one pole and one zero in each ear if I recall correctly) that
simply boosts the highs to the near ear and cuts them to the far ear.
It is certainly possible to do better with more complex filters.

The frequency-dependent time difference introduced by these filters is
pretty small compared to the overall ITD, so can probably be neglected;
but you might want to consider it and decide depending on your needs.

As a function of head diameter, I'd suggest scaling pole/zero frequencies
inversely with head size, starting with our numbers and whatever average
head size you care to guess.

We do not presently have any other publications on this stuff, but more
info on the Sprockets can be found at http://dev.info.apple.com/games

--------------------- mailto:lyon@research.apple.com ---------------------
  Dick Lyon                                      Distinguished Scientist
  Apple Computer 301-3M                              Apple Research Labs
  One Infinite Loop                                phone: (408) 974-4245
  Cupertino  CA 95014                                fax: (408) 974-8414
--------------- http://www.research.apple.com/personal/lyon/ -------------

   & Visiting Associate, Caltech Computation and Neural Systems Program
------------------- http://www.pcmp.caltech.edu/~dick/ -------------------