A small implementation of the Spleeter stem separation model in PyTorch. Using this model, audio files can be demixed into vocals, instrumentation etc.

Install the package using pip3 install ., then run

spleeter-pytorch audio-example.mp3

to separate the example file. The output will be located in output/stems.

The non-FFT parts of the Spleeter model can be converted to Core ML, for efficient inference on macOS/iOS devices. To perform the conversion, run

./convert-to-coreml

The .mlpackage will be located under output/coreml.

Note: The converted model corresponds to the Separator module and still requires the consumer of the model to manually perform the STFT conversion as performed in the Estimator. This is due to Core ML not supporting FFT operations yet.

• Currently this is only tested with the 2stems model. Feel free to get one of the other models and test it on them.
• There might be some bugs, the quality of output isn't as good as the original. If someone found the reason, please open a pull request. Thanks.

• Original Spleeter by deezer
• Original spleeter-pytorch by tuan3w

MIT.