On: Deepjazz: AI-generated 'jazz'

[squeaky-clean]

This is the commenter's point. Because the current way of training the model is by comparing its own output to the Pat Metheny tune, it doesn't work once you add more than a single song.

Musicians do learn from each other, but then they learn how to play what they like, or what sounds good. To this model, "what it likes" is a 100% representation of 'And Then I Knew' . You could swap the target song for another, but not for multiple targets at the same time without reworking the logic.

[ThomPete → squeaky-clean]

Isn't that just a matter of expanding the feedback loop to include other things than just the music?

I.e. aren't the mechanics there and is primarily limited by the the size of the feedback loop?

I am genuinely interested in the answer.