[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: HRTF: azimuth to ITD

Dear John,

This is a very important question, and lead to a very fundamental question:
what is the ITD? There are a variety of opinions on the matter going around,
concerning difference in time of arrival or time delay for the most
coherence between left/right signals. The addition of frequency variations
is also an important questions. 

Can you actually say there is sense in a measure ITD between 2 ears at 80
Hz? The wavelength a this frequency is enormous, human localization ability
being negligible, and finally the majority of HRTF measurement systems don't
have decent response at these low frequencies (as well as the rooms
response), so many (like CIPIC) simulate the results. 

At higher frequencies, the measured ITD values, and contour pattern (spatial
function one could say) vary greatly depending on the method used to
calculated the ITD. For example, with the max(ITD) azimuth varying from 60°
on front to 120° to the rear. 

The problem, as such, is both a signal processing and psycho-acoustic. While
many mathematical methods can be derived to calculate a time-difference,
there are in fact very few studies on the correlation between perceived ITD
and measured ITD. The few that have been done have focused on the frontal
plane, where there are less  differences between methods. 

While I realize that this response doesn't really answer your question, I
think it is important to highlight that in fact here is not a real consensus
on the issue, even though many take the idea of ITD as a given and simple

We are currently working on a publication for which much more information
will be available shortly. Regrettably, I can not diffuse the results before


-Brian FG Katz 

Brian FG Katz, Ph.D
Perception Située
BP 133
F91403 Orsay
tel. (+33)
fax. (+33)
e-mail Brian.Katz@xxxxxxxx


Date:    Thu, 7 Feb 2008 13:24:12 -0500
From:    JOHN WOODRUFF <woodruff.95@xxxxxxx>
Subject: HRTF: azimuth to ITD

i am doing some work that involves head-related transfer functions.  i have
been using the MIT and CIPIC databases and have noticed some unexpected
results regarding interaural time delay.  for example, in the case of the
MIT database (full,L) at 0 degrees elevation, the time delay associated with
the HRTF of 70 degrees azimuth is smaller than that of 65 degrees azimuth
for 80-150 Hz.  similarly, for 6.5-8 kHz, the time delay for 70 degrees
azimuth is larger than that of 75 degrees azimuth.  

i have noticed similar trends in the compact MIT database and in the CIPIC
database.  i can imagine that HRTF measurements taken from humans may have
slight discrepancies due to subtle head movements by the subject, but can
anyone explain why measurements using KEMAR would have non-monotonic
functions relating azimuth and ITD?  does what i'm seeing reflect physical
time delay or is it possibly an artifact from the measurement process or
signal processing involved?  

thanks very much, john

John Woodruff
Perception and Neurodynamics Lab
Ohio State University