Demucs
Separate vocals, drums, bass and other stems from any song.
- Star 数
- 3.4K
- Fork 数
- 314
- 未关闭 Issue
- 6
- 最近提交
- 2026-08-31
数据来自 GitHub API · 更新于 2026-10-15
Demucs 是什么？
Demucs (Hybrid Transformer Demucs) is a state-of-the-art music source separation model. One command splits a track into vocals, drums, bass and other — or just vocals vs. accompaniment for karaoke.
主要功能
- Four-stem and two-stem separation
- Six-stem model with guitar and piano
- CPU and GPU support
系统要求
- 平台
- Windows · macOS · Linux
- 显卡
- Optional; NVIDIA speeds things up
- 显存
- ~7 GB for default settings on GPU
- Python
- 3.8+
如何安装 Demucs
根据项目 README 整理。开源项目更新很快，请以 GitHub 页面上的最新说明为准。
pip
- 1Install.
python3 -m pip install -U demucs - 2Separate a song (outputs to ./separated).
demucs song.mp3 demucs --two-stems=vocals song.mp3
优势
- Excellent separation quality
- Simple CLI
不足
- Original Meta repo is archived; maintenance is limited
Demucs 替代品对比
| 软件 | Star 数 | 许可证 | 平台 | 显存 |
|---|---|---|---|---|
| Demucs Separate vocals, drums, bass and other stems from any song. | 3.4K | MIT | ~7 GB for default settings on GPU | |
| Ultimate Vocal Remover Desktop GUI for vocal removal and stem separation. | 26.6K | MIT | — | |
| Whisper OpenAI’s open-source speech recognition and translation model. | 110.5K | MIT | ~1 GB (tiny) to ~10 GB (large) | |
| whisper.cpp Whisper in plain C/C++ — fast on CPUs, Apple Silicon and mobile. | 54.4K | MIT | — | |
| faster-whisper Whisper reimplemented on CTranslate2 — up to 4× faster with less memory. | 25.9K | MIT | — | |
| WhisperX Fast transcription with word-level timestamps and speaker diarization. | 24.6K | BSD-2-Clause | — |
常见问题
Demucs 免费吗？
是的——Demucs 是开源软件，采用 MIT 许可证。商用前请确认软件及模型权重的许可条款。
Demucs 支持 Windows、Mac 和 Linux 吗？
Demucs 支持：Windows, macOS, Linux。
运行 Demucs 需要什么配置？
Optional; NVIDIA speeds things up. VRAM: ~7 GB for default settings on GPU
如何安装 Demucs？
1. Install. 2. Separate a song (outputs to ./separated).
Demucs 有哪些好的替代品？
常见替代品包括 Ultimate Vocal Remover, Whisper, whisper.cpp, faster-whisper，详见本页对比表。