Spleeter
Separate a mix into two, four or five estimated stems.
SeparationOpen-source code
- Maru integration
Maru extension available
Runs as a Maru browser module. Supported checkpoints and host requirements are documented on the extension page.
- Start here
- Use it as a straightforward separation baseline. Compare methods on the same source rather than treating more stems as better quality.
- Training data
- Deezer's privately assembled music collection; no downloadable training corpus is supplied with the library.
- Training rights
- The software release does not grant access to Deezer's source recordings or rights in the music you separate.
- License & output terms
- Repository: MIT; review the selected pretrained model release.
- Limitations
- An established baseline with separation artifacts and legacy TensorFlow dependencies; results vary by instrument and mix.
- Technique
- Spectrogram masking with U-Nets
- Runtime
- Local offline processing, with CPU and GPU configurations.
- Parameters
- Not itemized for this release/family. Parameter sizes shown elsewhere may exclude encoders, decoders or other runtime components.
Sources reviewed . Training and rights notes summarize provider disclosures. A model license does not clear every source recording or generated output. Parameter counts refer to the named variant or component, not total runtime memory.