Performance with PCS

[Ludvig-Joborn]

Hi again!
I'm wondering if you have considered evaluating MP-SENet with Perceptual Constrast Stretching? (PCS, see https://github.com/RoyChao19477/PCS)
I've seen it improve many models' performance on PESQ and COVL, so I'd be very interested in seeing how much the performance of MP-SENet improves as well.

Thank you!

[yxlu-0102]

Hi, I noticed that the PCS has been applied to the SEMamba (AN INVESTIGATION OF INCORPORATING MAMBA FOR SPEECH ENHANCEMENT), which used a similar architecture to MP-SENet.