[Smt-talk] Aesthetics of Computer-Generated Music

alex alex at slab.org
Sat Apr 16 13:50:25 PDT 2011

As Victor and several others have pointed out, "computer-generated
music" is a misnomer.  We write poems using English, but don't call
them English-generated.  Computer languages are not natural languages,
and act as a meta-medium, but nonetheless they are languages designed
for human expression. This makes them more than tools in by view, but
does not as yet grant them human-like agency.

Humans can, and frequently do compute.  As Eliot notes, and Allan
Paivio has demonstrated, we do more than this.  Most importantly to
music in my view, we dance.  Stockhausen recommended that every music
student should dance for three to four hours a week, and until digital
computers have better models of the analogue world we inhabit, I think
they won't be great musicians alone.

That said, in some areas electronic computers can compute faster and
more accurately than us, so it's a great idea to externalise some of
our human processes to them, such as the generation of discrete
musical patterns.  Musical development is much more difficult, for
example one of the more often stated difficulties for autonomous
generative music is how to program algorithms to stop, in a satisfying

Live coding practitioners don't need to encode musical development, so
are free of these problems.  Instead, for them software development is
musical development.  Andrew Sorensen's "Study in Keith" is a much
celebrated live coding example (takes a while to make sound):

I very much like this one from Scott Hewitt and Sam Freeman too:

More information and links to further examples are here:

Best wishes,



More information about the Smt-talk mailing list