Can you please share the complete training scripts ?

[yugeshav]

It will be good for the researchers / students if you share the working version of training script

[YangangCao]

to be finished...

[yugeshav]

ok, any timeline you are thinking ??

[henryxrl]

Do you know where the PHM from the paper comes in? I'm having a hard time understanding it... Also, how to reconstruct the audio signal from the output of the network?

[YangangCao]

@henryxrl Hi, I have no idea about your question. However, the network is totally same as the author of paper( I confirmed by email with author), maybe you can get some tips from network.

[amirpashamobinitehrani]

following @henryxrl 's comment, there is seems to be a Phase-aware β-sigmoid mask, attempting to separate the source, reverberated and noise parts of signal. I can't seem to find that in the implementation. Do you think that is part of the training step?