[sdiy] Phase shifts and instantaneous frequency

[Rainer Buchty]

On Wed, 16 Jul 2008, cheater cheater wrote:

> To answer: I said 'ears doing fourier transform' makes no sense simply
> as that: because it did not for me. Fourier analysis is a numerical
> tool that allows you to change the form of a signal in a very certain
> way. The human brain's psychoaccoustics may or may not display similar
> behaviour under certain conditions, but it is nearly impossible to
> prove that the way the human auditory system works is anywhere near
> frequency-oriented.

Oh, you mean all these hearing aid manufacturers working on 
frequency-dependent hearing aids and implants are on the wrong track?

Tell you something: how the ear detects and processes frequencies is 
understood rather well. Your daily dose of MP3 makes use of that 
knowlege, for instance.

The model of the ear can be roughly seen as a series of bandpass 
filters; unfortunately, the signalling behind can either transport fine 
frequency resolution or high volume, not both, which is why louder 
frequency domains silence neighbored ones w/ lower signal levels. (And 
I'll happily pass you the slides of our institute's dept. which deals 
with that kind of stuff.)

What the *brain* makes out of that is something completely different. 
It's not even too well understood what those intermediate nerve nodes do 
on the way from ear to brain -- what seems to be, though, is that 
permanent tinnitus typically is a product of those "inner" nodes, i.e. a 
(somewhat permanent) glitch in the signal processing behind.

Rainer