** Next:** Self-similar Synthesis
** Up:** The Ubiquitous Noise
** Previous:** Recreated Results
** Contents**

This section has tried to touch upon a few different issues concerning
noise. In general we view signals as either random or deterministic.
If a signal is not periodic and has an infinite amount of energy and
all its maximum and minimum values are in a finite range, with our
present state of signal processing we must treat the signal as
random. However, if the signal has finite energy (and
a finite number of discontinuities) we will be
able to mathematically, rather than statistically, derive and specifically
apply the Fourier transform theorem (Dirichlet
conditions[48, page 84]) to the signal. A signal
lives on the border of these dichotomy. The high frequency energy
of a
spectrum is finite, while the high frequency
region of a spectrum is infinite. The power spectrum of
random processes is usually also divided into two sections, a high frequency
region with a slope steeper than and close to ,
and the low frequency region which is flat. The flat low frequency region
implies that there is no long-term correlation in the signal, while the
steep high frequency slope implies a short-term correlation.
Keshner[20] points out:
*The presence of noise in MOSFET's, down to the lowest
frequency allowed by the limited observation time, suggests that the
division into just two subsystems is inappropriate.
*

The noise is an evolutionary signal, meaning that its whole past
history effects is present and future state. This implies a certain
type of memory in a process.
Dodge[9] finds fractals and noise
to be an interesting paradigm for computer-aided composition. He also
suggests that the ``memory'' of noise can account for its success.

The study of music as a noise has a certain value, in that it
treats a musical signal as a physical signal. The uniformity that a
model of music suggests exists on all levels of our perception
down to about 5 Hz. There are no psychological
issues to be considered. This is not to undermine the psychological
implication of music, but rather to suggest that if we would like
to make comments about music in a scientifically rigorous paradigm, it
is possible, as we really should, to ignore all psychological issues
(the most important of all of them being the assumption of ``intelligence'').
The study of music as noise assumes no intelligent entity except
the music itself.

** Next:** Self-similar Synthesis
** Up:** The Ubiquitous Noise
** Previous:** Recreated Results
** Contents**
Shahrokh Yadegari
2001-03-01