Fetching the paper…

TF-GridNet: Making Time-Frequency Domain Models Great Again for Monaural Speaker Separation · Around