Re: HRTF Question

[Sergio Rodríguez Soria <sergio@xxxxxxxxxx>]

Dear Chris,

As far as I am concerned, there is not a particular study about the
influence of the torso in the HRTF when the head is moving or when all
the body moves together. But, in fact, the HRTFs will be different... the
question is whether this difference will have a great perceptual impact
(of course the head movement HAS a great perceptual impact by itself). As
you should know, Algazi and Duda made several geometrical models for the
torso (the snowman, an elliptic torso, numerical analysis, etc) and they
concluded that the Torso may provide (but DON'T necessarily provides, it
depends on the individual, on the type of sound event, etc) cues for
elevation localization ONLY far from the median plane. The simple snowman
model could be added to an HRTF system in order to improve the elevation
localization performance in low frequencies, always problematic. This was
done in:

V.R. Algazi, R.O. Duda, D.M. Thompson, "The use of head-and-torso models
for improved spatial sound synthesis", Proc. 113 AES Convention,
2002, preprint 5712

D.Zotkin, J. Hwang, R.Duraiswami, L.S.Davis, "HRTF personalization using
anthropometric measurements", Proc. 2003 IEEE WASPAA, pp 157-160.

By intuition, given the symmetry of an spherical torso, the reflection
provided by it won't depend on the head position related to the
torso position (at least when the head rotates in azimuth), it will only
depend on the source position. Then, the entire HRTF will just change for
the new relative positon of the sound source. Maybe if the head rotates in
elevation... but it probably won't be significative. Anyway, this is only
for the snowman.

Other studies including torso influence are:

G.F. Kuhn, R.M Guernsey, "Sound pressure distribution about the human head
and torso" JASA, 73, 1983, pp. 95-105

K. Genuit, "A model for the description of out-ear transmission
characteristics", PhD thesis, Reinisch-Westfaelischen Technichen
Hochschule Aache, Germany, 1984

C.P. Brown, R.O. Duda, "A structural model for binaural synthesis", IEEE
Transactions on Audio Processing, 6(5), 1998, pp. 476-488

A good summary if provided by Algazi et al. in:

V.R. Algazi, C. Avendano, R.O.Duda, "Elevation Localization and
Head-related Transfer function analysis at low frequency", JASA 109 (3),
2001, pp 1110-1122.

Maybe you should see also:

P. Mackensen, "Auditive Localization, Head Movements, an additional cue in
Localization", PhD Thesis, Technische Universitat Berlin, 2004.

David,

Regarding to the question about "Generic HRTFs", maybe you should read an
interesting paper of Martens where several concepts about HRTFs are
discussed, like Individual, Customized, Averaged and Generalized HRTFs.
Following his definition, a generic public set of HRTFs would be the
MIT KEMAR one, and the CIPIC one could serve for customized HRTF systems.
See:

W. Martens, "Perceptual Evaluation of Filters Controlling Source
Direction: Customized and Generalized HRTFs for Binaural Synthesis",
Acoustical Science and Technology, 24 (5), 2003, pp 220-232


My best,

Sergio






On Tue, 16 Aug 2005, Chris Share wrote:

> Date: Tue, 16 Aug 2005 15:44:35 +0100
> From: Chris Share <cshare01@xxxxxxxxx>
> To: AUDITORY@xxxxxxxxxxxxxxx
> Subject: HRTF Question
>
> Hi,
>
> I'm currently doing some research into head-related transfer functions.
>
> In the articles that I've looked at so far, it's mentioned that the
> torso plays a role in the creation of the perceptual cues that enable
> sound source localization.
>
> I'm curious as to whether there has been any research comparing
> whole-body rotation (i.e. where the head and body move as one unit) as
> opposed to head-only rotation (where the body remains fixed as in
> sitting in a chair and turning the head to look sideways).
>
> It seems that the HRTFs for these two cases would be different as the
> ears' positions with respect to the shoulders/torso is not the same?
>
> Is that correct?
>
> Are there any articles that examine this issue?
>
> Cheers,
>
> Chris Share
>