What do we want to end up with?

From TheoryOrg
Jump to navigation Jump to search

Well, there are really a couple of options I can be happy with at this point.

1) The most difficult would be an interesting (in terms of sound) map to music. Once the text has been converted, a searching and pattern recognition algorithm would localize and extract interesting sounds, which can be thoght of as analogies for interesting ideas. This would be applied over numerous interviews and the results compared. Repetative patterns between files would be isolated and extracted to a new file and mapped back to language. Ideally, this would be where the different interviews crossed over the same point/idea. We would be left with three distinct objects:

  a) a sonification of the original text (many of these will be generated)
  b) a compilation of repeated patterns of (a), also a sonification
  c) a text generated by (b)

There one main problem with (1), that is coming up with a mapping which is also a bijection. We need this kind of function in order to map back to English and have it retain its grammatical structure that it went into music with (ie. so it makes sense). I can explain 1-1 and onto better, but briefly, for every unit of English there _must' be a sound and it must be 'unique_.

Now with a very simple map, the bijection would present little problem, but it also wouldn't be very musical. I think it would sound more like a collection of tones, which is unwanted. So,

2) Alternatively, we could create a map to music from English which focuses on composing music vs.(1). I really like the idea of using artist interviews as a source for the creation of music however, it isn't required for this kind of project. Creating the map in itself would be challenging enough, coming up with a decent algorithm(s?) that could repetedly transcribe text into complex sounds. Again, I think it would be fun to feed Naked Lunch through it and see what came out. :). We could extend this option to incorporate part (b) of number one, but I don't think it is entirely necessary.

Now, what are your ideas? -r


Last edit: Tue, 8 Oct 2002 08:18:53 -0700
(adsl-63-206-213-144.dsl.snfc21.pacbell.net)
Revisions: 1