RVC

Backend Space

HuggingFace Space

If inference fails with a 503 error, the space is sleeping. Open it on HuggingFace.co to wake it up, wait ~30 seconds, then try again.

Text-to-Speech — RVC NEW

Text Input

Language

Speed

1.0

Fetches audio from Google Translate TTS, converts to WAV, then auto-sets it as the Input Audio for RVC.

Text Input

Voice

Pitch

1.0

Uses your browser's built-in TTS (Edge voices on Microsoft Edge). Plays audio and records from microphone simultaneously — ensure your mic can pick up speakers, or use a loopback device.

Mic required

Model Source

Repository

Quick presets:

Folder

Model (.pth)

Index file (.index) — optional

Model file (.pth)

Click or drag .pth file here

Model weights file

Index file (.index) — optional

Click or drag .index file here

Improves voice similarity

Model URL (.pth)

Index URL (.index) — optional

Input Audio

Click or drag audio file here

.wav · .mp3 · .ogg · .flac · .m4a

Parameters

Transpose (semitones)

0 = no pitch change. +12 = one octave up. Use +12 for male-to-female, -12 for female-to-male.

F0 Extraction Method

rmvpe crepe crepe-tiny harvest dio pm

rmvpe is recommended. pm / dio are fastest. crepe is slowest but accurate.

Feature Retrieval Ratio

0.75

How much the index file influences the output timbre. 0 = disabled.

Filter Radius

If >= 3, applies median filter to reduce pitch artifacts (harvest/dio only).

Resample Audio Output

Target sample rate of the output file. 0 = same as model.

RMS Mix Rate (volume envelope)

0.25

0 = match output loudness to input. 1 = use converted audio loudness.

Protect Voiceless Consonants

0.33

Protect breath and unvoiced sounds from conversion artifacts. 0.5 = fully disabled.

Extra Options

Split Audio Into Chunks

Helps with very long audio or low memory.

Autotune Output

Apply pitch quantization after conversion.

Output Download

Debug log

Ready.

Credits

Backend Spaces / Creators

NeoPy

rvc-zero — primary inference space

huggingface.co/spaces/NeoPy/rvc-zero

ArkanDash

rvc-genshin-impact space

huggingface.co/spaces/ArkanDash

xTHExBEASTx

ultimate-rvc space

huggingface.co/spaces/xTHExBEASTx

Luminia

rvc-beatrice-voice-conversion space

huggingface.co/spaces/Luminia

Voice Model Repositories

PGR-RVC

Wuthering Waves RVC v2 — JP voices (Rover, Calcharo, Jianxin, Yinlin, Changli, Carlotta, Roccia, Camellya, Zani)

PGR-RVC/Wuthering_Waves_RVC_v2 — JP folder

TLME

RVC-Umamusume — Uma Musume Pretty Derby JP voice models

TLME/RVC-Umamusume

ArkanDash

Genshin Impact RVC models

huggingface.co/ArkanDash/rvc-genshin-impact

RVC (Retrieval-based Voice Conversion) is an open-source project. All voice models are community-made. Respect original voice actors and use responsibly.