Data format for training new model.

[mesielepush]

I've been trying to train a new model, but either midi.midi_to_sample() raises: "Detected multiple distinct time signatures." or try: samples = midi.midi_to_samples(path) raises an exception error. Is there a special configuration or preparation for the midi files?

[HackerPoet]

"Detected multiple distinct time signatures" is just a warning that it's skipping that midi file. You can ignore it assuming you still have samples that do work. The nature of the algorithm relies on measures and I used a fixed time step for simplicity, so I had to skip files with more than 1 time signature.

[User-71]

I have also found that when attempting to train a new model, I have only received the exception error. How can this be solved and what could cause this?

[AkuiChan]

This error is causing my code ot skip all the songs... and im unsure how to solve this issue