Furina EN - RVC v2 48k
English Furina voice conversion model for RVC, trained from clean in-game English voice assets.
Furina VO Dataset: /NaruseShiroha/Genshin-Furina-English
RVC Project (ver.2.3.260718): /RVC-Project/Retrieval-based-Voice-Conversion-WebUI
Model
- Architecture: RVC v2
- Sample rate: 48 kHz
- F0 guidance: Yes
- Training F0 extraction: RMVPE
- Selected checkpoint: Epoch 60
- Training clips: 977
- Training audio duration: 136.01 minutes (after filtering)
- Intended use: Voice conversion / covers
Clips shorter than 1 second and files with incompatible audio formatting were excluded from training, hence the lower clip count compared to the dataset.
Epoch 60 was selected after comparing checkpoints from epochs 50–100.
Recommended parameters
- F0 method: RMVPE or RMVPE+ if available
- Index rate: 0 works well; index is optional
- Protect: 0.33
- RMS mix: 0.5
happy cuddling
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support