README.md

---
title: RVC Voice Conversion
emoji: 🎙️
colorFrom: purple
colorTo: indigo
sdk: gradio
sdk_version: 6.9.0
app_file: app.py
pinned: false
license: mit
---

# 🎙️ RVC Voice Conversion

A **clean, minimal** HuggingFace Space for **Retrieval-Based Voice Conversion**.  
Record or upload audio → pick a model → convert in seconds.

## Features

| | |
|---|---|
| 🎙️ **Microphone** | Record directly in the browser |
| 📁 **File upload** | WAV · MP3 · FLAC · OGG · M4A and more |
| 📦 **Custom models** | Upload any RVC `.zip` (`.pth` + optional `.index`) |
| ✅ **Built-in model** | *zetaTest* auto-downloaded on first start |
| ⚡ **GPU / CPU auto** | Automatically uses CUDA if available |
| 🎚️ **Advanced controls** | Pitch · F0 method · Index rate · Noise reduction · Autotune · Split audio |

## Usage

1. **Convert tab** → record or upload audio
2. Choose a model (zetaTest is pre-loaded)
3. Set pitch if needed (e.g. `+12` to go up an octave)
4. Click **🚀 Convert Voice**

## Upload Your Own Model

Go to the **Models tab** and upload a `.zip` containing:
- `model.pth` — RVC model weights (required)
- `model.index` — FAISS index for better timbre matching (optional)

Compatible with models from [weights.gg](https://weights.gg) and [HuggingFace](https://huggingface.co/models?search=rvc).

## Credits

- Engine: [Ultimate RVC](https://github.com/JackismyShephard/ultimate-rvc) by JackismyShephard  
- Default model: [zetaTest](https://huggingface.co/megaaziib/my-rvc-models-collection) by megaaziib