Chris Hornbeck wrote:
On Sun, 30 Jan 2005 15:21:10 -0800, Bob Cain
wrote:


It is the bit depth that and sample rate together that
controls time resolution. Divide the sample period by the
number of discrete levels, (65536 for 16 bit) to get
approximately the time resolution. I think 16/44.1 at about
.35 nsec jitter is finer than the ear can detect.


This is very counter-intuitive. The extreme case of
infinite word length with no sample timing uncertainty
and no signal level modulation of sample point would
have, by implication, infinite bandwidth.

Infinite precision in time resolution is not at all the same
as infinite bandwidth. Even with infinite precision, the
Nyquist requirement of bandlimiting to half the sample rate
applies. Quantization jitter is just another form of noise.
I think it is more correlated with the signal than is
quantization noise but I'm not sure about that.


Wouldn't the reconstruction filtering process invalidate
this? Or is the answer over my head? Pretty likely.

Not sure why you would think that.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein