RE: Percussion!

> By the way, I've always thought that algorithmic composition (like Western
> music generally) has focussed WAY too much on "melodic" music,

I think it's tried, but has yet to succeed.  There's a chap at Sussex
who's resubmitting his PhD in which he has achieved some very interesting
results around somekindof jazz impro - high praise from Courtney Pine.
I always thought Pine was a machine, so not sure what that's worth.

> and the next
> person who wants to fiddle with algorithmic composition should aim for a
> program that sounds like Babatunde Olatunji rather than Mozart.

Yes, but let's at least make sure it's musical in some sense.
Reich would be relatively easy, don't you think?

> Altho that might involve a bit of thinking, because it's my
> impression that
> it's easier to teach a machine what intervals are "right", than to teach
> how to syncopate "right".

The interesting part is surely that it involves 10+ times more thinking
that programming (depending on programming language, of course: thanks
again for the libraries).  But of course you know that, which is why
you paremphasise right.  Yep: easy to make a little logic map for
interval changes, done a hundred times.  More interesting is the
relationship of wavelength and harmony, and what are rhythms but
wavelengths?  The interesting bit, I think, is making it do what is
not right - dissonance, off-beat, breaking and broken waves.

> (And, who knows, doing Mozart could be a good warm up to doing Olatunji.)

I'm not lowering myself to Mozart just yet.

> Cf. what Philip Glass said about patterns in Shankar's composition:
>   http://freshair.npr.org/dayFA.cfm?display=day&todayDate=09%2F21%2F1999
>     http://www.npr.org/ramfiles/fa/19990921.fa.ram
>     http://www.npr.org/ramfiles/fa/19990921.fa.rmm

Glass is, of course, also a computer, though running Algol *and* Forth.

> (Altho someone once sent me a program that turned rat genome data into
> ragtime music; that was pretty zippy!)

That is real intelligence for you.

Goddard
(Pronounced "Facile")