source file: m1414.txt
Date: Wed, 13 May 1998 09:35:18 +0300 (IDT)

Subject: Re: TUNING digest 1413

From: Robert C Valentine <bval@iil.intel.com>

> 
> just a really useless comment- the amount of dissonance in a sampled sound
> on the computer is inversely related to the percent of compression that a
> zipping utility can put on it.
> 
> -pete b
> 

Actually, it is the amount of noise, or randomness in a sampled sound 
(or any signal) which is inversely to its "zip"-ability. 

But while on the topic, the sample rate and number of bits in the sample
have a lot to do with this as well.

For instance, with infinite sampling precision, a sine tone whose
frequency is at an irrational division of the sample rate, would be 
uncompressable by a "zip" utility, since the same amplitude would never
repeat. 

With rational ratios, the amount of compressability would be given by the
simplicty of the ratio of the frequency to the sample rate. For instance,
a frequency of 1/2 the sample rate will begin to repeat after two samples,
but a frequency of 1/3 the sample rate will repeat after three samples.
Is a sine tone at 1/3 the sample rate more dissonant than one at 1/2 the
sample rate?

A real DSP or information theory person could go into greater detail,
and actually my main disagreement was with posing an equivalence between
"noise" in a signal and "dissonance" of a signal. Dissonance is subjective,
and noise can be measured, via the amount of redundancy in the signal
(which is the compressability).

Bob Valentine