Dannynis

Same here as of this date.

Perhaps i misunderstood the paper but doesnt the discriminator suppose to updated as well in the second atep?

hey, thank you for paying interest in this repo, I made it mostly for educational purpose, Im working on better version, that will use faster TDNN, meanwhile you could switch...

when commenting this code: ''' lmin = min(c.size(-1), spec.size(-1)) spec, c = spec[:, :lmin], c[:, :lmin] audio_norm = audio_norm[:, :lmin*self.hop_length] _spec, _c, _audio_norm = spec, c, audio_norm while spec.size(-1) <...

Hey, what do you mean by embedding a exactly ? The embedding from the first dense layer ? Also ive chosen 20 frames from 100 to 120 randomly to extract...

hey, currently is not comfortable to extract it, im working on version that will be more continent, meanwhile you could try pyannote package