Style transfer is pretty cool tool as someone who likes to play around with sound design. It will be a lot of fun to drop this on parallel channels and blend it

Style transfer is pretty cool tool as someone who likes to play around with sound design. It will be a lot of fun to drop this on parallel channels and blend it together into choruses and new instruments. Right now the generation still sounds a lot like loop packs smashed together anyone could technically make. But it is practical for anyone who really only cares for that style of sound but do not themselves have the familiarity to do it. Now they can just say what they want and hit regenerate, skipping the latent feedback cycle of iterating with humans or sifting through song snippets. My opinion on this style of content is that ai generation is simply accelerating us to the inevitable end of generic digital content, it isn't really changing it. It just happens to be also the optimal interface for discovering and not just generation.