[Smt-talk] Aesthetics of Computer-Generated Music
Greenwich
green at webster.sk.ca
Fri Apr 8 22:58:24 PDT 2011
I created a computer program several years
ago designed to create tonal melodies about
30-40 notes long. It was even nominated by
Discover magazine (unsolicited by me) to be
entered (entrance fee waived by them) in a
contest they were conducting that year 1997. My
webpage about that program is at:
http://www.greenwych.ca/serend4.htm
I think you'll all find it interesting. I
describe the basic principle involved in
making the program (Example URLs below).
It cured me of my compositional "writers'
block" that I developed when composing
music. It took months to write the program,
but only an hour or so to satisfy my need.
Namely:
As I showed everyone who'd look, the
program composed melodies very fast,
indeed, by the gross per minute. My motto
was: "To carry these tunes, you'll need a
forklift." I never needed to use the program
again after that first hour.
Writer's block -- gone!!
In my original advert for it, I offered
everyone who bought a copy of the program
they'd would get a free set of sparkplugs
(for their forklift). Unfortunately, you need an
old Tandy-100 laptop to run them and a used-
car dealer mentality to sell them.
However, I agree with Morse and Grauer --
especially as Victor wrote:
"...not even the most abstrusely
mathematical computer program, is ever
"fully generated" by computer.... I start with
an algorithm that seems promising, I listen to
the result, and then modify it over and over
again until I get something that "speaks" to
me.... I have a feeling many composers of
computer music work in more or less the same
way."
In my own case, I didn't go over and over
on it again and again. Just a few changes
were often enough, because of my second
program (or a Part 2) into which I put the most
promising melodies created by the first
program (see the webpage above).
Below are four URLs from the many
hundreds of melodies it made. These three
still are single-line melodies, but carry the
"illusion" of tonal harmony due to the "Part 2
program." There are still rarely any two
notes at-a-time (unless I sustain some notes or
add a human-composed line of counterpoint
-- a la as Gounod did w/ Bach).
This first one is 95% all computer made.
http://www.greenwych.ca/ser-357a.mp3
My favourite example is below, with only a
total of 5 or 6 notes changed, and 2
"repeats." (All the tunes get a 2 or 3 note
cadence to end them tonally). In this piece --
taking me only 30 minutes to "finish off" -- the
single melodic line of music is retained, but
has the aural illusion of harmny (because the
highest notes -- originally as short
unaccompanied tones, are sustained by part 2
of the program (or me) to connect them -- but
even without that, the illusion of harmony is
still there -- a la the Bach Partidas for violin):
http://www.greenwych.ca/ser2d-fx.mp3
On this next one, I took the computer's own
melody mostly as is, and added a counterpoint
line (either below it or over it).
http://www.greenwych.ca/ser-8a.mp3
Need a new short composition? Give me 10
minutes or so!
The last one here, is 50% inspired from
several combined as-is computer melodies &
50% me (but still mainly "single-line" music
-- and maybe 10% unconscious plagiarism).
http://www.greenwych.ca/ser384.mp3
--Bob & Fink
(The name Fink is the computer's; Bob is MY name.)
musicologists, retired
www.greenwych.ca
"A person who won't read has no advantage
over one who can't read." --Mark Twain
======
More information about the Smt-talk
mailing list