MVSEP vs Open-Unmix

MVSEP separates audio into voice and music parts and offers tools for vocal removal, karaoke creation, and stem splitting. Its extraction options include vocals, instrumental, bass, drums, piano, guitar, and other listed stem types; it also provides speech isolation, noise and reverb removal, and audio upscaling. The service lists more than 100 AI models for vocals, instruments, speech processing, transcription, MIDI extraction, and voice generation. It is available on the web, Android, iOS, and Windows, as well as through an API for creating separations, retrieving results, checking algorithms and queue information, managing users, and accessing separation history. Exports include MP3, WAV, FLAC, and M4A, and batch processing is available. The free Unregistered and Registered plans each allow 50 daily separations, tracks up to 10 minutes and 100 MB, and one concurrent job. Registered users also get additional output formats, medium queue priority, no captcha, and limited API access. Premium uses prepaid credits, supports tracks up to 100 minutes and 1000 MB, and offers full API access. Standard models charge by audio minutes; Ensemble models multiply minutes, and every job costs at least one credit.

Open-Unmix is a PyTorch reference implementation for music source separation, aimed at researchers, audio engineers, and artists. Its ready-to-use models split pop music into vocals, drums, bass, and other instruments; a separate model, umxse, is for speech enhancement and was trained on the 28-speaker Voicebank+DEMAND corpus. Pre-trained choices include umxl, umxhq, umx, and umxse, with spectrogram variants also available. Each source model uses a three-layer bidirectional LSTM, a design the project says cannot run online or in real time. The code supports reproducing results, developing architectures, and adding data, while training is documented separately from the package. Install through pip or use the Docker container; pre-trained models can be loaded in Python or through torch.hub, and a Colab notebook supports online experimentation without local setup. The command line handles WAV, FLAC, and OGG, not MP3; stempeg can extend formats, including STEMS export. The software is MIT-licensed; umxl weights are non-commercial under CC BY-NC-SA 4.0.