Skip to content

Latest commit

 

History

History
51 lines (33 loc) · 1.83 KB

README.md

File metadata and controls

51 lines (33 loc) · 1.83 KB
Raw

tidal-rave

Live coding RAVE neural audio synthesis models using TidalCycles and SuperDirt.

Dependencies

https://github.com/elgiano/nn.ar

Note: if you have a previous version of rave-supercollider installed in your SuperCollider Extensions folder, delete it, otherwise you will experience Symbol not found errors for libtorch.

Usage

  • Add the RAVE OSC target (rave.hs) to your TidalCycles Boot.hs.
  • Run rave.scd after SuperDirt has started.
  • Test the patterns in rave.tidal.

Models

You can find various CC-licensed models here: https://huggingface.co/Intelligent-Instruments-Lab/rave-models

You can also train your own models, see https://github.com/caillonantoine/RAVE.

For faster training via transfer learning, see https://github.com/victor-shepardson/RAVE.

Limitations

This is a simple way of using RAVE in SuperDirt and is quite limited.

SuperDirt

  • RAVE models are mono.
  • One RAVE model is assigned to one orbit.
  • rave.scd replaces SuperDirt output buses of Orbits 0, 1 and 2 with it's own bus.
  • No DirtPan currently interated into rave.scd.

Tidal

  • Most of rave.hs is only necessary because of rave.scd. If the latter can be improved, then only the custom parameters would be needed in Tidal.
  • Not possible currently to pattern the RAVE model itself due to the SuperDirt limitations described above.
  • RAVE latents not currently implemented on Tidal side.

TODO

  • Improved integration with SuperDirt
  • RAVE Prior variations
  • Mixing and match encoders and decoders
  • Patterning model names from Tidal (via Ndef?)

Discussion

https://scsynth.org/t/experiences-with-rave-ugen/6398/32

https://club.tidalcycles.org/t/tidal-x-rave-realtime-audio-variational-autoencoder/4973