# sigsep/open-unmix-pytorch

Open-Unmix - Music Source Separation for PyTorch

Repository: https://github.com/sigsep/open-unmix-pytorch
Canonical: https://ross.abutalabs.com/products/open-unmix-pytorch
Homepage: https://sigsep.github.io/open-unmix/
Language: Python
License: MIT
License Family: permissive
Last push: 2024-06-17T23:15:26+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2668, "days_push": 807, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1503, forks 206 (observed 2026-08-28T04:04:54.592721+00:00)

## What it is
Open-Unmix is a PyTorch deep learning reference implementation for music source separation, with pre-trained models that split pop music into vocals, drums, bass, and other stems. It is designed as a research baseline with reproducible results on the MUSDB18 dataset.

## Use cases
- separate a song into vocals drums bass and other stems
- remove vocals from a music track
- extract instrument stems from audio files
- baseline model for music source separation research
- train custom source separation models on own data
- speech enhancement with pretrained umxse model

## When to choose
- you need a well-documented, peer-reviewed reference implementation for source separation research
- you want ready-to-use pretrained models for stem separation in Python/PyTorch
- you want reproducible results on MUSDB18

## When to avoid
- you need real-time or low-latency separation in production
- you want the absolute state-of-the-art separation quality rather than a baseline
- you need a non-PyTorch framework or a GUI tool

## Facets
- artifact type: library
- maturity: maintenance
- function: audio-processing, machine-learning, deep-learning
- domain: machine-learning, deep-learning
- platform: python
- tags: music-source-separation, pytorch, stem-separation, audio, pretrained-models

## Member repositories
- sigsep/open-unmix-pytorch (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:54.592721+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:32:50.326658+00:00, confidence not recorded.
  - readme: https://github.com/sigsep/open-unmix-pytorch (fetched 2026-08-28T04:04:54.592721+00:00, sha 9134091c8e7b)
  - homepage: https://sigsep.github.io/open-unmix/ (fetched 2026-08-29T11:37:37.650597+00:00, sha 4a449a9d0bd5)
- Data as of 2026-08-30T08:39:29.467469+00:00.