Demucs banner

Machine Learning in Linux: Demucs – music source separation

Last Updated on March 6, 2023

Help message

usage: demucs.separate [-h] [-s SIG | -n NAME] [--repo REPO] [-v] [-o OUT]
                       [--filename FILENAME] [-d DEVICE] [--shifts SHIFTS]
                       [--overlap OVERLAP] [--no-split | --segment SEGMENT]
                       [--two-stems STEM] [--int24 | --float32]
                       [--clip-mode {rescale,clamp}] [--mp3]
                       [--mp3-bitrate MP3_BITRATE] [-j JOBS]
                       tracks [tracks ...]
Separate the sources for the given tracks

positional arguments:
  tracks                Path to tracks

options:
  -h, --help            show this help message and exit
  -s SIG, --sig SIG     Locally trained XP signature.
  -n NAME, --name NAME  Pretrained model name or signature. Default is
                        mdx_extra_q.
  --repo REPO           Folder containing all pre-trained models for use with
                        -n.
  -v, --verbose
  -o OUT, --out OUT     Folder where to put extracted tracks. A subfolder with
                        the model name will be created.
  --filename FILENAME   Set the name of output file. Use "{track}",
                        "{trackext}", "{stem}", "{ext}" to use variables of
                        track name without extension, track extension, stem name
                        and default output file extension. Default is
                        "{track}/{stem}.{ext}".
  -d DEVICE, --device DEVICE
                        Device to use, default is cuda if available else cpu
  --shifts SHIFTS       Number of random shifts for equivariant
                        stabilization.Increase separation time but improves
                        quality for Demucs. 10 was used in the original paper.
  --overlap OVERLAP     Overlap between the splits.
  --no-split            Doesn't split audio in chunks. This can use large
                        amounts of memory.
  --segment SEGMENT     Set split size of each chunk. This can help save memory
                        of graphic card.
  --two-stems STEM      Only separate audio into {STEM} and no_{STEM}.
  --int24               Save wav output as 24 bits wav.
  --float32             Save wav output as float32 (2x bigger).
  --clip-mode {rescale,clamp}
                        Strategy for avoiding clipping: rescaling entire signal
                        if necessary (rescale) or hard clipping (clamp).
  --mp3                 Convert the output wavs to mp3.
  --mp3-bitrate MP3_BITRATE
                        Bitrate of converted mp3.
  -j JOBS, --jobs JOBS  Number of jobs. This can increase memory usage but will
                        be much faster when multiple cores are available.

Pages in this article:
Page 1 – Introduction and Installation
Page 2 – In Operation and Summary
Page 3 – Help Message

Subscribe
Notify of
guest

This site uses Akismet to reduce spam. Please read our FAQ before making a comment.

0 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments