RE: [microsound] microsound and mathematics

[Adam Pierce <adam@xxxxxxxxxxx>]

Thanks for your comments jhno, I've been lurking on this list for months but
talk about GA has engaged my interest.

> the real solution is to move part or all of the burden of selection into
> the algorithm. if you can come up with an analysis technique that lets the
> computer guide the ga, then you can let it run overnight if it needs to.
> analysis is tough, though. whatever you do will have as much influence on
> the material as the generative algorithm itself (genotype -> phenotype
> step). i have dozens of ideas on the back burner, hope to have some
> interesting results to offer some day instead of just a theoretical
> perspective.

That's exactly what I've been thinking about. The problem is that music is
pretty much completely subjective. How do you write an algorithm which can
tell if something sounds good or not ?

I've had a couple of ideas along these lines:

1. Distribute the task. Make some sort of internet application in which
sounds are sent out to an audience & each audience member scores them from 1
to 10. If you can get 100s or even 1000s of listeners, you've got genuine
human selection in a large enough quantity to be useful.

2. Give it a library of music as a reference. Successful reproduction is
based on the similarity between the generated music and the library. You'd
have to be careful with this technique otherwise you'd get convergent
evolution. Perhaps this could be overcome by regularly changing the contents
of the library.


            __                              __
 _______ __/  \____________________________/  \__ _______
 \_____ |  \ o \       Adam Pierce        / o /  | _____/
   \___ |   \ o \________________________/ o /   | ___/
     \__|____\ o   www.doctort.org/adam   o /____|__/
           \__\____________________________/__/
               ~~~~~~~\____________/~~~~~~~
                         \______/
                            \/