README.md

---
title: Demucs Music Source Separation and Mixing
emoji: 🎵
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.42.0
app_file: app.py
pinned: true
thumbnail: >-
  https://cdn-uploads.huggingface.co/production/uploads/643999a70989db052b6257dc/GnmHlbzLDJP3dSsY_Al4s.png
---

# Demucs Music Source Separation and Mixing

This Space uses the Demucs model to separate music into individual stems (vocals, drums, bass, and other) and allows you to mix them back together.

## Features

- Separate audio into four stems: vocals, drums, bass, and other instruments
- Choose from multiple Demucs models for separation
- Select which stems to include in the final mix
- Option to save the output as MP3 with adjustable bitrate
- Real-time progress updates with a sleek interface

## How to Use

1. Upload an audio file (WAV format recommended)
2. Select a Demucs model from the dropdown menu
3. Choose which stems you want in the final mix (vocals, drums, bass, other)
4. Optionally, select "Save as MP3" and adjust the bitrate
5. Click "Process" to start the separation and mixing
6. Wait for the process to complete (you'll see real-time updates)
7. Download the mixed audio and individual stems

## Models Available

- htdemucs
- htdemucs_ft
- htdemucs_6s
- hdemucs_mmi
- mdx
- mdx_extra
- mdx_q
- mdx_extra_q

## Technical Details

- Built with Gradio
- Uses Facebook's Demucs for audio separation
- GPU-accelerated when available
- Supports both WAV and MP3 output formats

## License

This project uses the Demucs model, which is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License. Please refer to the [Demucs repository](https://github.com/facebookresearch/demucs) for more details on usage and restrictions.

## Acknowledgements

This Space is powered by the [Demucs](https://github.com/facebookresearch/demucs) model developed by Facebook Research.