Skip to content

2.0 Pre release

Pre-release
Pre-release
Compare
Choose a tag to compare
@CNChTu CNChTu released this 31 Jan 05:23
· 142 commits to v2.0_dev since this release
2d9964b

Diffusion SVC v2.0 is coming soon.

This model is a combination of NaiveV2, NaiveV2Diff, and Vocoder.
NaiveV2 and NaiveV2Diff is a cascaded training LYNXNet front stage and LYNXNet diffusion model.
They are extremely small in size and highly efficient.

You can train such a model by using configs/config_naivev2diff_comb.yaml and combining them with a fine-tuning vocoder using combo.py.

fine-tuning vocoder :https://github.com/openvpi/SingingVocoders