Where in your gwern link is the input format being crucial discussed? I couldn't find it.

I don’t know if gwern realized how powerful his model was. His examples are underwhelming, because you have to prompt it in a certain way to get it to generate chords. He was showing me samples and they were neat, but boring.

One day he posted something that sounded pretty amazing, and I was blown away. “More like that, please.” It had chords in it.

He didn’t pursue it past that. I did. So it’s possible that no one is aware of how crucial the input format actually is to the success of the music that I was able to produce.

(And “produce” is a fair description here — choosing the instruments was really important, and the model didn’t do it. It wasn’t as easy as press a button. It felt like I was suddenly a 15x music producer, since I made all those tracks in one night. Such is the power of ML.)

Do you have a write up anywhere with samples? I'd love to hear some of the better examples you have. I agree that most of what's out there is underwhelming

All of them are here: https://on.soundcloud.com/d6rX9goHHrSH3sQY9

Unfortunately I didn’t do a writeup (which gwern has given me a hard time about over the years, and he’s quite right!), so I have nothing to offer beyond those songs as a finished product. Maybe one day I’ll try to resurrect it for devs.

This is the best generated music I've ever heard. Are you blogging about your project? Where can I read more about your methods?

Sadly no, and nowhere. And with my primary focus being gamedev for the foreseeable future, the only way I see it being resurrected is if I need some generated music. That’s fairly low on the priority list for now, but it might preempt other things. https://github.com/shawwn/noh

To be immodest for a moment, my work serves as an example that it’s possible to do it, and better than anyone else, long before they figure out how. Many examples of this pop up throughout history, and I am gratified to be a small but real one.