Nonlinearity in excitation pattern modeling

[SOREN BUUS <BUUS@xxxxxxx>]

On 11-JUN-1997 13:03:25.60 Jont Allen wrote:


For some reason Soren's email really gets screwed up when I read it
via netscape. Does somebody know how to solve this problem?

One point I would like to add to Soren's email. It is essential to us=
e
a nonlinear cochlear model if you want to match JND and masking data.
With out it, the upward spread of masking and two-tone suppression
cannot
be matched in a realistic way. If you dont get masking right, then yo=
u
can
expect to get loudness right either, as loudness depends critically o=
n
the
masking.

Jont

--------------------------------
Dear Jont,
I agree with you about getting the masking
right and using a non-linear model.  For excitation-
patterns, the level dependence of the auditory filters
must be included.  Otherwise it won't work.=20
Using Zwicker's old excitation patterns does
the same thing.  In my mind, either approach is
a simple-minded way of taking into account the
cochlear non-linearity.  I like to keep things
simple, so I would tend to think that a full
cochlear model is overkill.  Moreover,  Peter
was looking for a model that might be believed
by a few more people than its originator.  It
seems to me that there is no universal agreement
about what is a good model of the cochlear
mechanics.  Thus, basing the excitation patterns
on a cochlear model might, to some extent, defeat
the purpose.

Best wishes,

S=BFren