Question about extra parameters to enhancement.py when using pre-Trained checkpoints

[RyanMetcalfeInt8]

Hi! When using enhancement.py with the pre-trained models from here: https://github.com/sp-uhh/sgmse?tab=readme-ov-file#pretrained-checkpoints, is it okay (recommended) to run with default parameters? Or do I need to set various parameters (like N, snr, t_eps, etc.) differently for each model?

I do see that the 2nd dereverberation model listed gives a recommended value for N, snr (--N 50 --snr 0.33) -- I just wasn't sure if there are similar recommendations for the rest of the models?

Thanks!
Ryan

[Huanianss]

May I ask how you are running the results metrics?