If you have a sufficiently recent version of macOS, you can just run
python3 -m pip install --user -U demucs
# Then anytime you want to use demucs, just do
python3 -m demucs -d cpu PATH_TO_AUDIO_FILE_1
# If you have added the user specific pip bin/ folder to your path, you can also do
demucs -d cpu PATH_TO_AUDIO_FILE_1
If you do not already have Anaconda installed or much experience with the terminal on macOS, here are some detailed instructions:
- Download Anaconda 3.8 (or more recent) 64-bit for macOS:
- Open Anaconda Prompt in macOS
- Follow these commands:
conda activate
pip3 install -U demucs
# Then anytime you want to use demucs, first do conda activate, then
demucs -d cpu PATH_TO_AUDIO_FILE_1
Important, torchaudio 0.12 update: Torchaudio no longer supports decoding mp3s without ffmpeg installed. You must have ffmpeg installed, either through Anaconda (conda install ffmpeg -c conda-forge
) or with Homebrew for instance (brew install ffmpeg
).