RemixIT: Continual self-training of speech enhancement models via bootstrapped remixing

Open paper
Benchmark Model Rank Results
speech-enhancement-on-deep-noise-suppressionSudo rm -rf (U=32)#13PESQ-WB: 2.95SI-SDR-WB: 19.7
speech-enhancement-on-deep-noise-suppressionRemixIT (w Sudo U=32)#19PESQ-WB: 2.34SI-SDR-WB: 16.0

Rank counts only results with a code link.