Demucs
State-of-the-art music source separation model by Meta for splitting tracks.
About
Demucs by Meta Research is a music source separation model that splits a song into drums, bass, vocals, and other stems. Version 4 uses a Hybrid Transformer architecture and reached top results on the MDX benchmark, running on GPU or CPU. The original repository is no longer maintained, with continued work on a community fork. Released under the MIT license.
Reviews (0)
Leave a Review
No reviews yet. Be the first to review!
Details
- Category
- Music & Audio Generation
- Price
- Free
- Platform
- Local/Desktop
- Difficulty
- Easy (2/5)
- License
- MIT
- Added
- Apr 3, 2026
Related Tools
PyTorch library for deep learning research on audio generation including MusicGen and AudioGen.
Audio generation framework by Meta including MusicGen for text-to-music.
Audio diffusion model by Harmonai for generating music samples.
Music generation model using masked acoustic token modeling.
Real-time music generation using Stable Diffusion on spectrograms.
Fast music generation model producing full songs with lyrics in seconds.