Does anybody reading this have prior experience with GANs/machine learning?
EDIT:
Currently reading
EDIT 2:
And this, which produces better results:
Unfortunatley both the links above are operating on MIDI data. I don’t think neural networks/machine learning had advanced nearly as far in the audio domain as it has in the visual. It’s a shame.
I’m thinking that simply setting up the tracks we have in a rotation on an icecast server is probably a more realistic goal.
But I need a little help understanding why this is preferable to simply listening on Bandcamp. Can someone help me out with that?
EDIT 3:
OK, there’s some raw audio as input here, and it’s actually pretty good. This is worthy of a deeper dive.
EDIT 4:
Seems worth a try:
@_mark if you can put the collected tracks somewhere I can get to them, I am willing to use them as training data for a neural network…