open-mmlab / Amphion

Amphion (/Γ¦mˈfaΙͺΙ™n/) is a toolkit for Audio, Music, and Speech Generation. Its purpose is to support reproducible research and help junior researchers and engineers get started in the field of audio, music, and speech generation research and development.
https://openhlt.github.io/amphion/
MIT License
4.45k stars 379 forks source link

Update DiffComoSVC #135

Closed Lokshaw-Chau closed 7 months ago

Lokshaw-Chau commented 7 months ago

✨ Description

[Please describe the background, purpose, changes made, and how to test this PR]

This PR updates the code of consistency distillation for SVC. It updates comosvc_trainer.py for more efficient training and hyperparams for consistency distillation for higher one-step generation quality. To test this PR, you can follow the guidelines at egs/svc/Diffcomosvc/README.md or check the demo at onedrive

🚧 Related Issues

None

πŸ‘¨β€πŸ’» Changes Proposed

πŸ§‘β€πŸ€β€πŸ§‘ Who Can Review?

[Please use the '@' symbol to mention any community member who is free to review the PR once the tests have passed. Feel free to tag members or contributors who might be interested in your PR.] @RMSnow

βœ… Checklist