Hey, you're always asking me where I find stuff...
Tuesday, March 21, 2006
The Future of Recombinant Music
So using a palette constructed entirely of sampled sound (even all the songs in the world!) performers can simply sing their compositions into existence. Now you can do that anyway of course - if you sing you are creating. But you're limited to only expressing one voice - monophonic. This technique can create a full polyphonic sound.
I think the more interesting aspect of this though is the possibility for automated music creation. In Neuromancer (? - maybe it was Count Zero) William Gibson described a kind of dub soundtrack that continuously recombined without human intervention. It would be really interesting to kick off the sCrAmBlEd?HaCkZ system with a short phrase and then feed the output back into the system again. I suspect one of two things would happen. The first could be that the feedback would build up very quickly and blow the speakers. The second could be that the music would change incrementally over time, so that after a while the music would be substantially different from what was originally performed.
Oh, there's a third possibility as well - that the music would never change at all. I guess the way around that would be to add a small amount of signal processing at the point where the output gets fed back into the system so as to introduce a certain level of randomness into the input.
I'd love to try it though - make sure you watch the video.
tags: music | sampling | sCrAmBlEd?HaCkZ
Gramophone records, magnetic tapes, vinyl records, digital samplers and computers have already liberated the samples long ago. But still - to infringe copyrights - one has to decide which sample one actually wants to steal. One has to arduously load audio files into sample editors or sequencers. One has to cut, copy, paste and arrange. All that takes precious creative energy and a lot of time.This guy has created a system called sCrAmBlEd?HaCkZ that takes input from a performer (e.g. a singer), analyzes the input, breaks it into segments, matches the segments against a database of previously sampled segments of music, and then plays back the matching segments, all in real time.
Enough of that!
So using a palette constructed entirely of sampled sound (even all the songs in the world!) performers can simply sing their compositions into existence. Now you can do that anyway of course - if you sing you are creating. But you're limited to only expressing one voice - monophonic. This technique can create a full polyphonic sound.
I think the more interesting aspect of this though is the possibility for automated music creation. In Neuromancer (? - maybe it was Count Zero) William Gibson described a kind of dub soundtrack that continuously recombined without human intervention. It would be really interesting to kick off the sCrAmBlEd?HaCkZ system with a short phrase and then feed the output back into the system again. I suspect one of two things would happen. The first could be that the feedback would build up very quickly and blow the speakers. The second could be that the music would change incrementally over time, so that after a while the music would be substantially different from what was originally performed.
Oh, there's a third possibility as well - that the music would never change at all. I guess the way around that would be to add a small amount of signal processing at the point where the output gets fed back into the system so as to introduce a certain level of randomness into the input.
I'd love to try it though - make sure you watch the video.
tags: music | sampling | sCrAmBlEd?HaCkZ
Comments:
Post a Comment
Links to this post:
- RSS