Re: spatialisation using headphones

["Ward R. Drennan" <wdrennan@xxxxxxxxxxxxxxxxxxxxxxxx>]

hamish,

Is it a good idea? That I don't know, but if you generate a testable
hypothesis, test it avoiding pitfalls and artifacts, and your results add
something new to what's known about cross-modal attention, then it's
a good idea. A key is developing a knowledge base of what's known
and not know about cross-modal attention and addressing a
specific, tractable problem.

question #2. Using loudspeakers in an anechoic chamber is a good
idea, but getting a really good anechoic chamber is difficult. The
double-walled booths aren't really anechoic. You could measure
reverb times. Also, it might be useful to determine what effect reverb
has on bimodal attention. There's so many things that could be
studied.

Using personalized HRTFs is a second option, but that's time
consuming and would require training with people experienced at
doing that (see Fred Wightman). There is an AUDIS catalogue of
HRTFs which gives I think 10 HRTFs which encompass a variety of
headsizes. You can buy it (from France, I think) for a modest fee. I
can't seem to find it on the web, but someone on the list may know
how to get it.

Headphone HRTFs that are not specific to a particular listener tend
to give the impression of an image lateralized inside the head rather
than an externalized acoustic percept. Such an effect may or may
not be good enough for your purposes. It seems that speakers in a
room with minimal reverb might be the best/easiest option.

Another item of note is this-- pinna spectral cues matter at high
frequencies-- i.e. greater than 4KHz and usually in the range of 6-
10KHz, so if you used lower frequency stimuli it's not likely that
those cues would be of great concern.

Ward Drennan
> Hello list!
>
> I'm a PhD student at the Brain Sciences Institute in Melbourne, about
> to embark on my first experiment looking at cross-modal attention
> affects on event-related brain potentials.  I'm planning on using
> simulated "spatial" stimuli in one of my tasks.  I have some flashing
> lights located on the left, right and center in front of a subject.
> There are some beeps that I'd like to play thrpough headphones AS IF
> they were coming from the location of the flashing lights.
>
> They don't have to be exactly localised at the location of the lights,
> but I would like them to be vaugely from the same area.  What I'm
> looking for is a way of "spatialising" my (already made up) sound
> stimuli.  I'm aware that the HRTF for each person is different, but am
> think that perhaps i could use a small set of typical HRTF's that
> would cover "large-headed", "small-headed" etc people.
>
> What I was wondering was
>
> 1) is this a good idea
> 2) where should i start looking for ways of actually implementing
> this?
>
> I have a passing familiarity with MatLab etc
>
>
> many thanks for any pointers in the right direction!
>
> -hamish-


Ward Drennan, Ph. D.
Kresge Hearing Research Institute
Phone: (734) 763-5159
Fax: (734) 764-0014