demucs
Music source separation in the waveform domain.
Decision gist · record as of 2026-08-14
Yes. Demucs is actively maintained, has no known vulnerabilities, low install friction, and permissive licensing. It delivers state-of-the-art stem separation out of the box. Install if you need to separate music into drums, bass, vocals, or other stems; skip if you only need metadata extraction or don't work with audio.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10+.
- torch installation may require system CUDA/GPU drivers if GPU acceleration is desired; CPU-only inference is supported but slower.
- Low friction: pure Python wheel with no compiled dependencies beyond torch.
License · maintenance · safety
MIT License (permissive) — MIT License permits commercial and private use with minimal restrictions; you may use, modify, and distribute Demucs freely provided you include the license notice.
last release 2026-07-11 (34 days) · last repo commit 2026-07-11 · 3,028 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 400,505 downloads/mo, #6,939 on PyPI
Alternatives
Verify before relying
pip install demucs
python3 -m demucs -n htdemucs_ft path/to/song.mp3
# Outputs separated stems to ./separated/htdemucs_ft/song/- Actual inference speed and memory requirements for typical song lengths on CPU vs. GPU hardware.
- Quality degradation or artifacts when separating non-Western music genres or atypical instrumentation.
- Whether the 6-source model (adding guitar and piano) has improved since the acknowledged piano source artifacts mentioned in the description.
What it is and what it does
Demucs is a music source separation model that breaks a single audio track into isolated stems—drums, bass, vocals, and other accompaniment. Version 4 uses a hybrid transformer architecture that processes both waveform and spectrogram representations simultaneously, achieving state-of-the-art separation quality (9.0 dB SDR on the MUSDB HQ test set). The model was trained on MUSDB HQ plus 800 additional songs.
You invoke it via command-line or Python API to process MP3, WAV, or other audio formats. It outputs separate audio files for each stem. The package includes multiple model variants: htdemucs_ft (fine-tuned, recommended), htdemucs (baseline), hdemucs_mmi (retrained), and an experimental 6-source model. Separation runs on CPU or GPU; GPU is significantly faster but not required.
Use it for
- Extract vocals from a song for karaoke or vocal-focused remixing workflows.
- Isolate drum and bass tracks for beat analysis, re-arrangement, or music production.
- Create instrumental versions by removing vocals for background music or licensing.
- Prepare training data for music information retrieval or other audio ML tasks.
- Analyze song composition by examining separated stems independently.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes.
Demucs is actively maintained, has no known vulnerabilities, low install friction, and permissive licensing. It delivers state-of-the-art stem separation out of the box. Install if you need to separate music into drums, bass, vocals, or other stems; skip if you only need metadata extraction or don't work with audio.
Install
demucs on PyPI
Before you install
Low friction: pure Python wheel with no compiled dependencies beyond torch. Actively maintained as of July 2026 with recent updates. Requires Python 3.10+.
Requires Python 3.10+. torch installation may require system CUDA/GPU drivers if GPU acceleration is desired; CPU-only inference is supported but slower.
License in practice
MIT License permits commercial and private use with minimal restrictions; you may use, modify, and distribute Demucs freely provided you include the license notice.
Quickstart
pip install demucs
python3 -m demucs -n htdemucs_ft path/to/song.mp3
# Outputs separated stems to ./separated/htdemucs_ft/song/Verify before relying
- Actual inference speed and memory requirements for typical song lengths on CPU vs. GPU hardware.
- Quality degradation or artifacts when separating non-Western music genres or atypical instrumentation.
- Whether the 6-source model (adding guitar and piano) has improved since the acknowledged piano source artifacts mentioned in the description.
Package facts
| License | MIT License permissive |
| Python support | Supports the current Python release >=3.10 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 10 packageseinopshuggingface-hubjuliuslameencnumpypyyamlsafetensorssphntorchtqdm |
| Maintenance | Actively maintained 34 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 400,505 / month, #6,939 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | License :: OSI Approved :: MIT LicenseTopic :: Multimedia :: Sound/AudioTopic :: Scientific/Engineering :: Artificial Intelligence |
Evidence: demucs-4.1.0-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “separate drums bass vocals”
- demucsDemucs separates music into individual stems—drums, bass, vocals, and…
- openunmixSeparates music into individual stems (vocals, drums, bass, other…
- audio-separatorSeparates audio files into multiple stems (vocals, instruments,…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also audio-separator · openunmix · torchaudio · encodec · Gammatone · aubio · snac · spotdl · noisereduce