Retrieval-based-Voice-Conversion-WebUI
RVC-Project/Retrieval-based-Voice-Conversion-WebUI · Python · MIT
The default WebUI for voice conversion covers
RVC WebUI is how many creators train a voice model and run covers. Heavy community, multilingual UI, and a permanent place in AI music workflows.
Score
Setup at a glance
git clone https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI.git Deep dive
RVC WebUI is the retrieval-based voice conversion tool that needs little introduction in the vocal-synth community: it converts one voice into another in real time while keeping the performance intact.
Its signature trick is top-1 retrieval — replacing source features with training-set features to eliminate timbre leakage — which is what keeps the output from sounding like a blend of both voices. Training is fast even on modest GPUs, and ten minutes of low-noise speech is enough for a decent model.
Model fusion (ckpt-merge) lets you blend checkpoints into new timbres, and the project ships a WebUI plus docs in Chinese and English; a ModelScope demo studio lets you try it without installing.