Time-Domain Audio Source Separation Based on Wave-U-Net Combined with Discrete Wavelet Transform.

Tomohiko Nakamura, Hiroshi Saruwatari

Browse the full ICASSP paper archive.