Retrieval-based-Voice-Conversion-WebUI

An easy-to-use voice conversion framework based on VITS - for MacOS

🔒 Security Update: This version includes PyTorch 2.6.0+ to address critical security vulnerabilities. See SECURITY.md for details.

FAQ (Frequently Asked Questions)

English | 中文简体 | 日本語 | 한국어 (韓國語) | Français | Türkçe | Português

The base model is trained using nearly 50 hours of high-quality open-source VCTK training set. Therefore, there are no copyright concerns, please feel free to use.

Please look forward to the base model of RVCv3 with larger parameters, larger dataset, better effects, basically flat inference speed, and less training data required.

There's a one-click downloader for models/integration packages/tools. Welcome to try.

Training and inference Webui

Real-time voice changing GUI

Features:

• Reduce tone leakage by replacing the source feature to training-set feature using top1 retrieval;
• Easy + fast training, even on Apple Silicon;
• Training with a small amounts of data (>=10min low noise speech recommended);
• Model fusion to change timbres (using ckpt processing tab->ckpt merge);
• Easy-to-use WebUI;
• UVR5 model to quickly separate vocals and instruments;
• High-pitch Voice Extraction Algorithm InterSpeech2023-RMVPE to prevent a muted sound problem. Provides the best results (significantly) and is faster with lower resource consumption than Crepe_full;
• Apple Silicon acceleration with MPS (Metal Performance Shaders) supported.

Check out our Demo Video here!

📦 Standalone macOS Application

New! RVC-MacOS is now available as a standalone macOS application bundle (.app) that can be distributed without requiring users to install Python or dependencies.

Download Pre-built App

Download the latest release from GitHub Releases:

1. Download RVC-MacOS-Installer.dmg (~500MB)
2. Open the DMG file
3. Drag RVC-MacOS.app to your Applications folder
4. Launch from Applications or Spotlight

Important - First Launch:

• First-time setup takes 5-10 minutes to download AI models
• The app will download ~1.5GB of required model files automatically
• You'll see clear progress messages during the download
• Internet connection required for first launch
• Subsequent launches are instant (models are saved)

What Gets Downloaded on First Launch

RVC requires AI model files to function. On first launch, the app automatically downloads:

• HuBERT base model (~189MB): Voice feature extraction
• RMVPE pitch models (~110MB): Pitch detection
• Pretrained RVC models v1 (~600MB): Voice conversion
• Pretrained RVC models v2 (~600MB): Enhanced voice conversion

Total: ~1.5GB downloaded once, used forever

Disk space needed: At least 3GB free (for models + temporary files)

Build from Source

To build the standalone application yourself:

# Quick build - creates ~500MB app bundle
./build_app.sh

# Create DMG installer
./create_dmg.sh

The build script will:

1. Check for Python 3.8-3.10 (required by fairseq)
2. Install dependencies
3. Build the .app bundle (~500MB without models)
4. Models will be downloaded by end users on first launch

The built application will be in dist/RVC-MacOS.app and the DMG installer in dist/RVC-MacOS-Installer.dmg.

For detailed build instructions, see PACKAGING_MACOS_PYTHON_APPS.md.

---

Environment Configuration

Python Version Limitation

It is recommended to use venv to manage the Python environment.

For the reason of the version limitation, please refer to this bug.

python --version # 3.8 <= Python < 3.11

MacOS One-click Dependency Installation & Startup Script

By executing run.sh in the project root directory, you can configure the venv virtual environment, automatically install the required dependencies, and start the main program with one click.

Quick Start (without models)

sh ./run.sh

This will start the application. If models are missing, you'll need to download them first - see "Automatic Download" under the Assets section below.

First-time Setup (with automatic model download)

sh ./run.sh --download-models

This will download all required models before starting the application. This is recommended for first-time setup and may take several minutes depending on your internet connection.

Manual Installation of Dependencies

1. Install pytorch and its core dependencies, skip if already installed. Refer to: https://pytorch.org/get-started/locally/

   pip install torch torchvision torchaudio

2. Install the required dependencies:

   pip install -r requirements/gui.txt

Preparation of Other Files

1. Assets

RVC requires some models located in the assets folder for inference and training.

Automatic Download (Recommended)

The easiest way to download all required models is using the included download script:

python download_models.py

Or use the run.sh script with the --download-models flag:

sh ./run.sh --download-models

Check/Download via Web Interface

By default, RVC can automatically check the integrity of the required resources when the main program starts.

Even if the resources are not complete, the program will continue to start.

• If you want to download all resources when starting the web interface, please add the --update parameter:

  python web.py --update

• If you want to skip the resource integrity check at startup, please add the --nocheck parameter.

Download Manually

All resource files are located in Hugging Face space

You can find some scripts to download them in the tools folder

You can also use the one-click downloader for models/integration packages/tools

Below is a list that includes the names of all pre-models and other files required by RVC.

• ./assets/hubert/hubert_base.pt

   rvcmd assets/hubert # RVC-Models-Downloader command

• ./assets/pretrained

   rvcmd assets/v1 # RVC-Models-Downloader command

• ./assets/uvr5_weights

   rvcmd assets/uvr5 # RVC-Models-Downloader command

If you want to use the v2 version of the model, you need to download additional resources in

• ./assets/pretrained_v2

   rvcmd assets/v2 # RVC-Models-Downloader command

2. Download the required files for the rmvpe vocal pitch extraction algorithm

If you want to use the latest RMVPE vocal pitch extraction algorithm, you need to download the pitch extraction model parameters and place them in assets/rmvpe.

• rmvpe.pt

   rvcmd assets/rmvpe # RVC-Models-Downloader command

Getting Started

MacOS (Recommended)

For first-time setup with automatic model download:

./run.sh --download-models

For subsequent runs:

./run.sh

Direct Launch

Use the following command to start the WebUI.

python web.py

To download all required models before starting:

python web.py --update

Credits

• ContentVec
• VITS
• HIFIGAN
• Gradio
• Ultimate Vocal Remover
• audio-slicer
• Vocal pitch extraction:RMVPE
  • The pretrained model is trained and tested by yxlllc and RVC-Boss.

Thanks to all contributors for their efforts

RVC-WebUI-MacOS

A macOS-optimized version of the Retrieval-based Voice Conversion WebUI, specifically designed for Apple Silicon (M1/M2/M3) Macs.

Features

• Voice conversion with high-quality results
• Easy automated setup with automatic model download
• Optimized for Apple Silicon (M1/M2/M3) Macs
• User-friendly web interface
• Support for various audio formats
• Real-time voice conversion
• Training capabilities for custom voice models

Requirements

• macOS 12.0 or later
• Apple Silicon Mac (M1/M2/M3)
• Python 3.8 to 3.10 (due to fairseq compatibility, see bug)
• 8GB RAM minimum (16GB recommended)
• 10GB free disk space

Installation

Option 1: Quick Setup with run.sh (Recommended)

1. Clone this repository:

git clone https://github.com/audiohacking/RVC-WebUI-MacOS.git
cd RVC-WebUI-MacOS

2. Run the setup script with automatic model download:

sh ./run.sh --download-models

This will automatically:

• Create a virtual environment
• Install all dependencies
• Download all required models
• Start the web interface

Option 2: Manual Installation

1. Clone this repository:

git clone https://github.com/audiohacking/RVC-WebUI-MacOS.git
cd RVC-WebUI-MacOS

2. Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate

3. Install dependencies:

pip install -r requirements/gui.txt

4. Download models:

python web.py --update

Usage

Using run.sh (Recommended)

./run.sh

Manual Launch

1. Activate the virtual environment:

source .venv/bin/activate

2. Start the web interface:

python web.py --port 7860

3. Open your web browser and navigate to:

http://localhost:7860

Training Your Own Model

1. Prepare your training data:

   • Place your audio files in WAV format
   • Recommended duration: 10-50 minutes of clean audio
   • Sample rate: 16kHz or higher
   • Place files in logs/your_experiment_name/0_gt_wavs/

2. Start training:

   • Use the web interface to start training
   • Select your experiment name
   • Choose training parameters
   • Click "Start Training"

Directory Structure

RVC-WebUI-MacOS/
├── assets/
│   ├── pretrained/    # Pretrained models (included)
│   └── rmvpe/         # RMVPE model files (included)
├── logs/
│   └── your_experiment_name/
│       ├── 0_gt_wavs/     # Original audio files
│       ├── 1_16k_wavs/    # 16kHz converted files
│       ├── 2a_f0/         # Pitch information
│       └── 2b-f0nsf/      # Processed pitch information
├── requirements/
│   └── gui.txt        # GUI dependencies
└── web.py             # Main application file

Troubleshooting

1. If you encounter "No supported Nvidia GPU found" message:

   • This is normal for M-series Macs
   • The application will automatically use MPS (Metal Performance Shaders)

2. If you get "address already in use" error:

   • Try using a different port: python web.py --port 7861

3. If model loading fails:

   • Verify file permissions
   • Check if the model files are present in the assets directory
   • Try reinstalling the dependencies

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Original project: fumiama-Retrieval-based-Voice-Conversion-WebUI
• Modified and optimized for macOS by Nevil Patel