A web application for music source separation based on the Open-Unmix project.
This application requires that your GPU be compatible with CUDA SDK version 9.0+. You can check whether your GPU can run the application by cross-referencing its microarchitecture here.
- Separates music sources (vocals, bass, drums, harmonics, other) from a given audio input
- Built using React, Django, and Open-Unmix
- Supports pyTorch with CUDA 12 for GPU acceleration
- Upload an audio file to the application.
- Choose the target model you want to use for separation.
- Click the "Separate" button.
- Wait for the separation process to complete.
- Download the separated audio files for each target source.