I was getting a bit bored of music AI videos, since they didn’t talk a lot about the inner mechanisms of the programs used and they often used the same methods, but this is Really interesting, great stuff!
You are providing us with outstanding content. Thank you for all the great hints that computer science can provide on the philosophy of art, through the lens of music
ironically I've recently been thinking about generating music for games by stitching together small pieces using a sort of Markov-chain-based automaton. this video shows another interesting process that could be expanded on for such things!
One of these things that I like messing around with is Musenet. It's more of a music continuer, so definitely not random, but I still think it's really interesting
How might one actually experiment with the instrumentation to go from the CA output to something audible that doesn't sound like an animal dying on a keyboard?
Seeing how computers can make this kind of music, I'm thinking what would happen if we give an AI a piano, 0 rules, and another AI to rate the music the first one made. What kind of musical style would it turn into?
I'm certainly not an AI expert, but my guess is it would be completely random noise. As I understand it, any sort of generative AI needs to be trained on past material, the style of which it replicates. If there's none of this material, the first AI would just generate stuff randomly (or not at all) and the second one would also discriminate at random. Every music-generating AI is trained on some set of pieces.
I was getting a bit bored of music AI videos, since they didn’t talk a lot about the inner mechanisms of the programs used and they often used the same methods, but this is Really interesting, great stuff!
You are providing us with outstanding content. Thank you for all the great hints that computer science can provide on the philosophy of art, through the lens of music
ironically I've recently been thinking about generating music for games by stitching together small pieces using a sort of Markov-chain-based automaton. this video shows another interesting process that could be expanded on for such things!
No promises, but there’s a good chance they’ll be a markov chain video at some point. I’ve been looking into it as well.
Look forward to this!
Enjoyed this a lot!
One of these things that I like messing around with is Musenet. It's more of a music continuer, so definitely not random, but I still think it's really interesting
we need more videos :)
Nice, but you should have said that this is not the only type of cellular automata. The cells can have more states, for example.
How might one actually experiment with the instrumentation to go from the CA output to something audible that doesn't sound like an animal dying on a keyboard?
🤯
Seeing how computers can make this kind of music, I'm thinking what would happen if we give an AI a piano, 0 rules, and another AI to rate the music the first one made. What kind of musical style would it turn into?
I'm certainly not an AI expert, but my guess is it would be completely random noise. As I understand it, any sort of generative AI needs to be trained on past material, the style of which it replicates. If there's none of this material, the first AI would just generate stuff randomly (or not at all) and the second one would also discriminate at random. Every music-generating AI is trained on some set of pieces.
cgm
It's not real AI if it isn't learning
don’t think he ever said ai lmfao
@@746kviews4 the guy just assumed that the video was about A.I XD
I think someone here didn't watch the video properly