Audio file preprocessing

[VlaDanilov]

Hello, thank you for your work, I'm looking forward to use it in my studies.
I have a short question, I don't have much experience in that filed, so I don't fully understand how audio should be preprocessed in order to use your losses? Should I use Mel spectrogram or FT? What those tensors, that you give as an input to your losses, are?
Thank you