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

Re: Headphone to ear canal Transfer Function

Dear Sergio,

think about it this way: if you are unlucky and the headphones you use
are not perfectly symmetric in their transfer function, then they will
introduce their own, (entirely unphysiological !) interaural spectral
pattern differences, and that would most definitely affect the fidelity
of your VAS. Most headphones have some "manufacturing tolerances", i.e.
the transfer function of the left and right ear transducer are not
exactly identical. Whether these asymetries are large enough to create a
problem is not something you'll know until you try, but because of these
potential problems, most serious researchers will calibrate their
individual set of headphones, and will work the head phone calibrations
into their virtual acoustic displays.
It is conceivable that specialist headphones, like the tube phones from
etymotic, have transfer functions that are so flat and reproducible that
you may not need to worry about this individual calibration, but with
"bog standard" headphones designed for domestic audio this is definitely
not the case. Also, etymotic headphones are "guaranteed flat" only for a
limited frequency range, and you would need to check whether your
signals remain within that range. HRTFs have quite complex spectral
patterns at the high frequency end, but the short wavelengths of high
frequencies mean that they can also easily be influenced by standing
waves / reflections  & interference effects. Calibrating a headphone
system for the high frequency end is therefore particularly tricky. In
other words, the more high frequencies you have in your signals, the
more you have to worry about getting the calibrations right, and if you
want to guarantee VAS fidelity all the way up to 20 kHz then your best
(perhaps only) option will be to rig up small microphone capsules that
you can insert in the ear canal of the subjects under the headphone so
that you can calibrate the headphones in situ, as it is done for example
in the Wightman / Kistler lab. Look at their papers for details.



Sergio Rodríguez Soria wrote:

Dear list members,

I work with HRTF models  and in the next months need to do some
listening tests in order to validate the models. Normally I don't take
into account the headphone to ear canal Transfer Function (which includes
the headphone transfer function). Is that very important for reproducing
virtual sound through headphones? If it is so, I will need the
headphones free-field transfer function (for compesating just the
headphones), and/or,  if there exists in public domain, the headphone to
ear canal transfer function (for compensating the headphones and the
transmission path between the headphone's output and the ear canal
entrance - given that the HRTF model reproduce the transmission between
the "free field" and the ear canal entrance). Do I am right? Does anyone
know about a public database for these kind of signals?

Besides of that, what headphones are recommended for these tests?

Thanks in advance

Sergio Rodríguez Soria
Graduate Student
Signal Processing Laboratory
University of São Paulo


Dr. Jan Schnupp
University Laboratory of Physiology             St Peter's College
Oxford University                               New Inn Hall Street
Parks Road, Oxford OX1 3PT, UK                  Oxford OX1 2PL
Tel (01865) 272513                              Tel (01865) 278889
Fax (01865) 272469