Usage

Choose any or all of the instruments in a track to isolate and remove the instrument from the track. Outputs the isolated instrument track and the rest of the audio in a merged track.

Expected runtime (after startup): 1 minute.

Algorithm

Demucs is a state-of-the-art music source separation model, currently capable of separating drums, bass, piano, guitar, and vocals from the rest of the accompaniment. Demucs is based on a U-Net convolutional architecture inspired by Wave-U-Net. The v4 version features Hybrid Transformer Demucs, a hybrid spectrogram/waveform separation model using Transformers. It is based on Hybrid Demucs (also provided in this repo), with the innermost layers replaced by a cross-domain Transformer Encoder. This Transformer uses self-attention within each domain, and cross-attention across domains. The model achieves a SDR of 9.00 dB on the MUSDB HQ test set.

Sample track

Song - Cobie Sample Artist - JBlanked Source - Free Music Archive License - CC BY-NC-ND