OpenAI MuseNet

  Quote

 

We’ve created MuseNet, a deep neural network that can generate 4-minute musical compositions with 10 different instruments, and can combine styles from country to Mozart to the Beatles. MuseNet was not explicitly programmed with our understanding of music, but instead discovered patterns of harmony, rhythm, and style by learning to predict the next token in hundreds of thousands of MIDI files. MuseNet uses the same general-purpose unsupervised technology as GPT-2, a large-scale transformer model trained to predict the next token in a sequence, whether audio or text.

https://openai.com/blog/musenet/

This is the best AI generated (or "co-composed") music I have heard until now, here is a sample track:

They are looking to collaborate with musicians and also general feedback, their e-mail is at the bottom of the page   

Yep that fooled me. It still kind of sounded a bit "soulless" but if you told me the guy was contracted to do that song for a company then I would of 100% believed it.

So reading up on it, it is essentially markov chains on steroids.

I bet this thing could churn out tech house without breaking a sweat.

This jazz song reminded me of The Warheads though. Maybe I am harsh, but after listening a while these guys sound like their stuff is put together by a neural network as well.

https://soundcloud.com/jrs-1-1

i look forward to all commercial radio just being replaced with computer generated sound-a-likes, hosted by Microsoft Sam AI Radio host. 

a good chunk of the chord voicings don't sound like choices humans would've made, at least in the context of that song. felt a touch rigid in 'play' but just programming in some slowing off the grid at key points could start to simulate human 'feel' and swing and such. definitely a lack of delicacy with velocity and such too. 

this is amazing