Open-Unmix

What Open-Unmix is

Open-Unmix is a free PyTorch reference implementation for music source separation aimed at researchers, audio engineers, and artists. Its ready-to-use models separate pop music into vocals, drums, bass, and other instruments. It also includes umxse, a speech enhancement model trained on the 28-speaker Voicebank+DEMAND corpus. Pre-trained choices include umxl, umxhq, umx, and umxse, along with spectrogram variants. The code is intended for reproducing results, developing architectures, and adding data. Each source model uses a three-layer bidirectional LSTM; the project states that this architecture cannot run online or in real time. The Python package installs with pip, and a Docker container is available. Models can be loaded from Python or torch.hub, and a Google Colab notebook supports online experimentation without local setup. The command line accepts WAV, FLAC, and OGG, but not MP3; stempeg can expand supported formats, including STEMS export. The software is MIT-licensed, but umxl weights are non-commercial under CC BY-NC-SA 4.0. Technical help is offered through GitHub issues, Gitter, and Google Groups.