Compute and memory efficient universal sound source separation

Open paper
Benchmark Model Rank Results
speech-separation-on-whamrImproved Sudo rm -rf (U=36)#7SI-SDRi: 13.5
speech-separation-on-wsj0-2mixSudo rm -rf (U=36)#22SI-SDRi: 19.5

Rank counts only results with a code link.