I am wondering if, like AARON, an artificial intelligence software that would be programmed to improvise acoustic piano solo, which I love, could do a similar task on playing music. I mean composing music with obeying and understanding orders such as minor = sad, slow, calm, etc... or program it on major that would be equal to joyful, fast, exciting, etc... No doubt if this does not exist yet, someone will invent it pretty soon. The unique factor, however, that would render it similar to AARON is that everytime it plays it would play a new piece, a new improvisation, so you would never hear the same tune twice. And on the other hand, doing this would be of any value in music? In visual painting, renewing is valuable for it stays, you can actually take a picture of every new production.
Just thoughts...
This AARON thing was so inspiring... Plus 30 years working on such a project, that's like the biggest challenge ever!...
I think it would be valuable to the music world if it were recorded perhaps. Personally, I would love to be able to hit a button on something (like my computer) and have music play (that still sounds appealing) that I know no one else has ever heard. But, if it's not being recorded, then who knows what the world is missing.
ReplyDeleteMusic was probably the first human art-form that was automatically generated by computer algorithms as soon as computers were able to synthesize sound. Even at a consumer level, before any algorithmically generated visual art was attempted, music generation was already in the hands of early Commodore 64 and Apple users.
ReplyDeleteFor many years now there exist many sophisticated programs that use probabilistic-logical programming, Markov Models (ask Asutosh about that)and other synthetic models to generate polyphonic music, sometimes in a particular author "style", like Mozart's or Bach's.
PRISM is one of such music generators. Check the link to see an example.