Fetching the paper…

Dual-path RNN: efficient long sequence modeling for time-domain single-channel speech separation · Around