[sdiy] jitter analysis

[Richard Wentk]

At 15:47 09/07/2004 +0200, Czech Martin wrote:

>As long as we model the ear as filter bank or DFT, I can not follow
>this argument.

Well, that's the problem. You're assuming it's a very simple filter bank 
that does a very simple spectral decomposition. But that's simply not true. 
The ear is actually highly nonlinear, especially at high sound levels. A 
pair of ears are capable of localising transients to a higher resolution 
than you'd expect from the standard 20-20k argument. And we know very 
little indeed about how timbral perception works. There may well be - and 
very likely is - a lot of psychoacoustic processing that relies on 
correlation/pattern recognition effects, as well as hunting for the 
fundamental and all kinds of other processes.

>A transient has a very broad spectrum, an ideal pulse is "white"
>in some way. So you can not say that a transient has this or
>the other frequency, but the spectrum is really broad in long time
>Fourier transform, or sharper and decaying in hopping short
>time transforms.

Indeed. But you don't hear a transient as a frequency either. That was 
never my point. What you hear is an event that starts at a certain point, 
has a certain timbre, and - if there's anything else going - fits into a 
rhythmic framework. That's actually a hell of a lot of processing and 
interpretation.

Anyway - the jitter issue should be easy enough to test empirically. If you 
sample at different frequencies, if it's pure hf jitter perceived by a 
simple hf recognition process, sooner or later the perception of fatness 
will disappear. If you can still hear fat when you're sampling at 4k, 
presumably something else is happening.

Someone out there with some really good converters might want to try this 
and let us know how they get on.

I think 'fat' is very likely a combination of processes. Aside from jitter 
I think it would be very useful to check something obvious like spectral 
plots to see what they tell us.

I also think it's interesting that everyone knows what 'fat' sounds like, 
but no one seems to have much of a clue what it correlates to in physical 
terms. :-)

Richard