Machine Learning in Linux: Demucs – music source separation
This series looks at highly promising machine learning and deep learning software for Linux.
Demucs is billed as “a state-of-the-art music source separation model, currently capable of separating drums, bass, and vocals from the rest of the accompaniment”. It’s based on a U-Net convolutional architecture inspired by Wave-U-Net. Version 4 features Hybrid Transformer Demucs, a hybrid spectrogram/waveform separation model using Transformers. It sounds impressive, but what are the results like?