Offline voice cloning implementation
Just finished https://github.com/Mimocro/supertonic-voice-cloning - simple project to run the black magic and train any voice (legally obtained of course).
Just put voice as reference.wav into data, copypaste commands from example and get trained json that can be used with supertonic 3 models without hitting a paywall (15$, really?)
I was already too late to buy the voice builder subscrubtion anyway and wouldnt pay that much for a .json file too but your project is really good. I cant really use it myself as I use RX 590 AMD card but really good man keep it up!
That’s great work, my friend. Right now, the audio output using your repo isn't quite crystal clear yet, but the vocal timbre is already very close to the original voice. I hope you keep up the good work.