[Smt-talk] Aesthetics of Computer-Generated Music

Eliot Handelman eliot at colba.net
Sat Apr 9 08:40:47 PDT 2011

Nicolas Meeus wrote:

 > What would be faked, in the case of computer-generated music, is its
 > intentionality. The extent to which such music could produce an
 > aesthetic effect strongly depends on the intentionality of the
 > listener.

The problem in autonomous composition, as I conceive it, is to
generate the kind of structure that invites intentionalistic
interpretations. Is there anything else that (human) composers
necessarily do? The rest might not be much more than a program note.

In my system, Jack & Jill, in progress for a decade, I wouldn't say
that intentionality is "faked": rather, it's theoretically motivated,
only its locus is not that of emotional response. It's about shape and
motion perception, expectancy realization and violation, a theory of
simplicity, and ideas about the influence of the visual system in
perceptions of formal direction ("motion asymmetries," an idea I'm
pushing as a "musical universal"). These form as it were the "bodily"
substrate of the (neurally) later metaphorization of that kind of
(worldly) perception in which the sense of music arises in a person.

The system has an explicit theory of intention, expressed in my slogan
"music listens." It's the idea that that a "call/response" structure
is very basic to music, and that the response seems to respond to the
call. Thus music is always responding to itself, and exactly in our
perception of this respondingness do we perceive something to be
music. At any rate that's my theory.

My earliest take was that composing music means that you hear and in
some way grasp the music that you are composing. This where average
computer-generated music usually fails: one gets the idea quickly that
the music is not at all aware of itself, does not act "meaningfully",
lacks the "right kind of structure," ie, that which invokes the
sense of an agentful response to itself.

This implies that a composing system must "be aware" (ie, have an
accessible representation) of what it has done. At one time I called
this "listening," but now I find this to be too loaded.  Currently, Jill 
is an
analytical dual of the composing system, able to take raw input
(midi), discovering in this relations of object and structure of the
kind that correspond to the representations of the composing system.

All of this is partly meant as a philosophical and cognitive theory of
what composing might imply. How does this reflect on actual practices?
It would imply that composers create structures, not intentions. The
T.S. Eliot story about how the poet "must have meant" what someone
found in his poem, is, I think, more real than it may seem.

-- eliot

Eliot Handelman PhD,

More information about the Smt-talk mailing list