unilight/seq2seq-vc

A sequence-to-sequence voice conversion toolkit.

53
/ 100
Established

Built on Transformer architectures with TTS-based pretraining, it enables prosody-preserving voice conversion through multiple approaches: non-autoregressive seq2seq with automatic alignment search (AAS), ground-truth-free foreign accent conversion, and the original Voice Transformer Network (VTN). Recipes follow ESPNet/Kaldi conventions and support both parallel and non-parallel training paradigms across datasets like CMU ARCTIC and LJSpeech.

108 stars.

No Package No Dependents
Maintenance 13 / 25
Adoption 9 / 25
Maturity 16 / 25
Community 15 / 25

How are scores calculated?

Stars

108

Forks

15

Language

Jupyter Notebook

License

MIT

Last pushed

Mar 15, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/unilight/seq2seq-vc"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.