Upload a single-track MIDI to the AWS DeepComposer service
npm i -g deepcomposer-upload
Input MIDI files should have a tempo that remains static. Only the first 8 bars will be processed.
deepcomposer-upload -i input.mid -n nameoftrack -m genre-rock-1 -o output.mid
deepcomposer-upload \
-i input.mid \
-n nameoftrack \
-m ar-cnn-bach \
-o output.mid \
--max-percentage-removed 100 \
--max-notes-added 50 \
--sampling-iterations 100 \
--creative-risk 1.5
The filename of the input MIDI file
The name of the sample, used to register in Music studio (required when uploading)
The ID of the model used to generate new tracks, defaults to genre-rock-1
genre-rock-1genre-pop-1genre-jazz-1genre-joco-1genre-symphony-1
ar-cnn-bach
The name of the output filename of the generated MIDI file, if you wish to save it
Autoregressive only
The maximum percentage of initial notes removed (0-100)
Autoregressive only
The maximum notes to be added (50-1000)
Autoregressive only
The sampling iterations (0-100)
Autoregressive only
As risk increases, compositions will become more experimental (0.5-6)
Lists tracks in the input MIDI file, and ignores uploading to the DeepComposer service
Selects the track number to process, as shown by --list-tracks