explr/spleeter repository overview

This image is derived from official spleeter image researchdeezer/spleeter⁠, and integrated with all official pretrained models⁠.

It simplified the command line parameters. To separate a music at <your_dir>/music.flac into vocals and accompaniment, run:

docker run --rm -v <your_dir>:/data explr/spleeter music.flac

It's also possible to use the command line parameters like -c:

docker run --rm -v <your_dir>:/data explr/spleeter music.flac -c mp3

or, to use the 5stems model:

docker run --rm -v <your_dir>:/data explr/spleeter music.flac -p spleeter:5stems

The default entrypoint is:

ENTRYPOINT [ "spleeter", "separate", "-p", "spleeter:2stems-16kHz", "-c", "flac", "-o", "/data", "-i" ]

but it seems like that the later params overrides the former.

To use the spleeter manually, you need to override the entrypoint and mount your working directory at /data.

docker run --rm --entrypoint /bin/bash -v <your_dir>:/data explr/spleeter

The pretrained model had been placed on /model.