These demos use the MusicVAE model to encode a multi-instrument
NoteSequence, and call sample() to
obtain a new NoteSequence that matches it. Each section uses a
different model checkpoint (melody, or chords) to generate the samples.
Note: This demo takes a couple seconds to complete.
Input:
Result: Loading...
It took:
It took:
Samples:
Input:
Result: Loading...
It took:
It took:
Samples:
Total Leaked Memory: