remember the deeplearning stuff where it'd take one painting's style and apply it to another one

when are we getting that for music, that'd have unlimited (meme) potential