Yossi Adi

I also got similar results. Couldn't improve above ~13.4 SDRi. Regarding the distorted samples you should set norm =True in the librosa write function.

Hi @ChuxiJ, Thanks for showing interest in our work! Unfortunately, it will take us a bit more time to release the training code, I can not really say when but...

@ChuxiJ, In case you meant something else than the training code, can you please elaborate? :)

Hi @sh-lee-prml, 1) These metrics consistent with human perception up to some level, they are definitely not perfect, but can give you a nice proxy for faster model evaluation. Anyways,...

Hi @tsaitaoguei, Did you manage to download the 34B model eventually?

Hi @liqiang28, Did you manage to solve this issue? It is not clear from your second comment. If not, can you please provide more details regarding the issue?