Re: electroacoustics - rap to tap to zap to frap

Subject: Re: electroacoustics - rap to tap to zap to frap
From: Eliot Handelman (
Date: Thu Aug 05 2004 - 19:55:34 EDT wrote:

>And yet, if all the computer is doing is analyzing prior works and
>mathematically determining new ones,
this is "all"? Do you have any idea how complex this is? Run any of the
programs Cope has made
public and you'll discover none of these produce even a coherent tune.
It's all just random rejumble
without any ongoiing analysis of what's taking place. The result is
exactly what you expect crummy-sounding
computer-generated slog to sound like. Then listen to the pices he says
were written with a "latger program." Ask yourself:
how much larger? Can you figure out those modules that figure out how
to generate music that might mike sense? This is
an amazing achievement, if it's bveing pulled off as advertised, which
we suspect it isn't. There is no music
theory around that says how to create something that seems add up, or
even recognize when
that takes place. Any program that generates music successful in this
way, especially if not restricted to a particular style,
involves major origina;l theorizations of music-structural or listening
process that no one else knows about. This
isn't anything to be taken for granted.

I've been working at such a program for quite a while. It does in fact
attempt to supply machinery to theorize about
what makes sense, and it's about to acquire a brain based on
machine-learning research. I find this
an exceptionally hard problem that has twisted my head around about
what music is
and I feel a great sense of regret knowing in advance that any results I
may yet get
will be thoroughly taken for granted by people who say "of course, it's
just a question of applying the
right mathematics." Musicians should know how complex it must be to
synthesize something
that adds up. We need much more wait-a-minute-ism.

-- eliot

This archive was generated by hypermail 2b27 : Sat Dec 22 2007 - 01:46:03 EST