[sdiy] Phase shifts and instantaneous frequency

[Aaron Lanterman]

On Jul 16, 2008, at 4:31 PM, cheater cheater wrote:

> it is nearly impossible to prove that the way the human auditory  
> system works is anywhere near frequency-oriented.

Uh... Fred Juang, Mark Clements, and Chin Lee all have offices down  
the hall from me, and Tom Barnwell and Ron Schafer had offices down  
the hall from me before that before they retired, and having spent a  
good combined 140 or years between them working on audio and speech  
processing and recognition, I think they'd be awfully surprised to  
find that "it is nearly impossible to prove that the way the human  
auditory system works  is anywhere near frequency-oriented."

It's way near frequency-oriented.

I've been on a good dozen PhD committees involving speech recognition  
or audio enhancement or things along those lines, and they all start  
with extremely well developed models of the ear that center around  
frequency domain concepts. Understanding those concepts is what makes  
cochlear implants work. Google "Fourier cochlea," without the quotes.  
Go to Amazon and search on "psychoacoustics," and get any text you  
see. These principles have been established for decades. There a lot  
of complexities and subtleties (many of which we've been debating on  
this list), and much that is still not understood, but if it wasn't  
for that frequency-orientation of your ear, we wouldn't have iPods.

> It might be oriented in any other of infinitely many possible  
> domains, not the time domain, not the frequency domain. Certainly  
> the ear does not have the facility to integrate numerically, much  
> less do something like the Laplace transform.

I don't want to sound harsh - please don't take my next statement like  
that - but there's really layers upon layers of misunderstanding there  
concerning how mathematics is used and can be useful in model physical  
systems. Mother nature may not be running trapezoidal integration  
calculations, but it solves Maxwell's equations in real time nontheless.

- Aaron