Hacker News new | ask | show | jobs
by jschveibinz 20 days ago
Just spitballing, but what if multiple AI agents played different parts and had to synchronize, harmonize and even improvise with each other? This might add the texture necessary to make the result more "listenable"?

Anyone out there trying to demonstrate this?

3 comments

All AI music I've heard sounds average, common "denominatorish".

That to me seems like a general observed behaviour with LLMs. Adding more agents I don't think it will break the mold.

On the other hand the majority of people listen to bland/generic music all the time. Thus I'm really kot that surprised when "AI artists" become popular.

I think it is less that the music is "listenable" and more that people want to listen to music made by humans.

I do not want to pay for AI generated music from a place like Spotify. If I wanted AI generated music I would just go to the AI music generator and pay them directly.

That is still slop. It's slop because it's generated by a computer. I don't care how technically proficient it is. The entire point of listening to music is that I am listening to a person communicate. Sure, they use computers to generate the sounds, but at a minimum they have the taste to put together the arrangement.

Maybe Spotify could focus on AI features that help artists discover audiences, organize events, and get paid.

Instead they just want to steal the very little money that artists actually make.