Re: [MUSIC-IR] Re: musical complexity

[Eliot Handelman <eliot@xxxxxxxxx>]

On 03/09/2011 9:32 PM, Kevin Austin wrote:
> On first hearing, these works (and others . . .) may appear complex 
> (and complicated), but through study and continued listening, it is 
> possible to hear through this initial response. Therefore, if 
> 'complexity' is susceptible to reduction through continued exposure 
> and education, I continue to suggest that the "complexity" is not in 
> the acoustical signal, but is a perceptual category. As such, it has 
> no metric.

The point of the metric is to measure what's being perceived and as you 
point out
even for a single individual there are different metrics that develop 
over time. But that doesn't matter,
because the metric could ask, "given time and exposure, what kind of 
hearing could evolve," leaving the
question only who this is supposed to be about.

IMO, this is currently a free-form theoretical issue, in which what 
matters is telling some sort of story
that goes from signal to perception.

I presented on a related issue at SMPC 11, in which I tried to show how 
you can infer the
sorts of musical objects that Bach thought about, through an analysis of 
difference in a couple
of collections. Slides are here:

http://www.cs.mcgill.ca/~asigle1/bachmemory.pdf

The point is, beginning with notes, what sort of structures can be built 
up that _arguably_ correspond
to what Bach was thinking about or hearing? Let's call this 
"bach-structure," and let's say that a piece has
"bach-complexity" when there are specific parallels to the kind of 
structure you'd get in Bach.

I actually did apply this idea to Temperley's computer-generated 
melodies, and it's demonstrable,
at one level of description anyway, that you do NOT get the 
bach-complexity in these computer generated diddlies.

Note this is not to say that anyone DOES hear in this way, including 
Bach: it's just a model, testable on the
kind of predictions it should make.

-- eliot