# deezer/spleeter

Deezer source separation library including pretrained models.

Repository: https://github.com/deezer/spleeter
Canonical: https://ross.abutalabs.com/products/spleeter
Homepage: https://research.deezer.com/projects/spleeter.html
Language: Python
License: MIT
License Family: permissive
Topics: deezer, python, tensorflow, audio-processing, pretrained-models, bass, drums, vocals, deep-learning, model
Last push: 2026-06-18T15:22:32+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 88, release rhythm 8, longevity 100
- inputs: {"age_days": 2533, "days_push": 76, "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 28402, forks 3060 (observed 2026-08-28T04:11:48.324112+00:00)

## What it is
Spleeter is Deezer's music source separation library with pretrained TensorFlow models that splits audio into stems (vocals, drums, bass, piano, other). It works as a CLI tool, a Python library, or a Docker image, and runs up to 100x faster than real-time on GPU.

## Use cases
- separate vocals from a song's instrumental
- extract drums and bass stems from an mp3
- make a karaoke/acapella track from any song
- split audio into 2, 4, or 5 stems
- train a custom source separation model on my own dataset
- batch process a music library into stems

## When to choose
- you need fast, pretrained music stem separation without training your own model
- you want a CLI or Python API for separating vocals/accompaniment
- you have a GPU and need faster-than-real-time separation

## When to avoid
- you need real-time live separation in a production audio pipeline
- you need separation of stems beyond the pretrained models (e.g. guitar-only) without your own training data
- you need a maintained, actively developed solution - the project is largely in maintenance mode

## Facets
- artifact type: library
- maturity: maintenance
- function: audio-processing, machine-learning, deep-learning, cli
- domain: machine-learning, audio, media
- platform: python, cli, cross-platform
- tags: source-separation, stem-separation, pretrained-models, tensorflow, music, karaoke, deezer, audio, docker, gpu

## Member repositories
- deezer/spleeter (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:48.324112+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 2026-08-29T16:54:23.133263+00:00, confidence not recorded.
  - readme: https://github.com/deezer/spleeter (fetched 2026-08-28T04:11:48.324112+00:00, sha dd7f9aa40a4f)
  - homepage: https://research.deezer.com/projects/spleeter.html (fetched 2026-08-29T07:50:55.116248+00:00, sha af76f8434b1a)
  - site_page: https://research.deezer.com/about (fetched 2026-08-29T07:50:55.125463+00:00, sha 781e64c3f8e2)
  - site_page: https://www.deezer.com/features (fetched 2026-08-29T07:50:55.127287+00:00, sha d9984361dcc8)
  - registry_pypi: https://pypi.org/pypi/spleeter/json (fetched 2026-08-29T07:50:55.132305+00:00, sha 5be635ac3bb0)
  - site_page: https://www.deezer.com/en/company (fetched 2026-08-29T07:50:55.129218+00:00, sha 15ae0972dc81)
  - site_page: https://www.deezer.com/en/channels/new (fetched 2026-08-29T07:50:55.130841+00:00, sha 782eab920e1a)
- Data as of 2026-08-30T08:39:29.467469+00:00.