music generation example

[Jonathhhan]

i tried to use this model for music generation: https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/audio/music_generation.ipynb#scrollTo=1mil8ZyJNe1w

its very rough yet, and not sure if everything works as expected.
maybe you can find something wrong...
especially my replacement for tf.random.categorical(pitch_logits, num_samples=1) could be wrong.
the model is very small, so it is included in the example...
https://github.com/Jonathhhan/ofxTensorFlow2/tree/music_generation_example/example_music_generation

[Jonathhhan]

I guess it works now...

[danomatika]

Without trying it, I would say we could connect it to a synth with ofxMidi or generate audio directly with ofxPd.

[Jonathhhan]

@danomatika yeah, I already connected it with midi out from ofxMidi. ofxPd would be nice, too. Often it sounds quite random, but I can also hear some musical structure. Have to play a bit with the training settings...

[danomatika]

Looks like you need to track the duration for each pitch and send the noteoff individually. That might give you better rhythm and structure on the output.

[Jonathhhan]

It is interesting, that the output seems to get more structure after running for some time.
But maybe that is, because I choose a more or less random sequence for the beginning...

[Jonathhhan]

I added velocity to the model (and added the edited notebook to the example): https://github.com/Jonathhhan/ofxTensorFlow2/tree/music_generation_example/example_music_generation_2
And actually wonder, how to implement the music transformer: https://colab.research.google.com/notebooks/magenta/piano_transformer/piano_transformer.ipynb