Description

I made a retrain of this one without using any of the pretrain models (such as TITAN) this time. Apologies for putting in a swear word in the title from my previous one. Dataset contains just a 1-second clip ripped from the episode "Desmond's Big Day Out" retrained without looping the audio to 1 minute because that causes the HiFi-GAN to mess up. *Recommended to keep transpose from a higher setting (e.g. 12 or higher)* *Recommended to set the search feature ratio up to 1.*