4 ms·
so how would the process of training a speaking AI go ? would you input the actor voice samples and subtitles from a movie, then train it till the output is sim
by slim 2y ago
so how would the process of training a speaking AI go ? would you input the actor voice samples and subtitles from a movie, then train it till the output is similar enough to the actors voice from the movie ?
- slim 2y agowhat test data would they use ?
- numpad0 2y agoJust couple minutes of data through 10-20 minutes of training with RVC WebUI[0] on included base model into VC Client[1] gets you to 90% there. But that's nearly an year old method, so I'm sure OAI has its own completely novel architecture for extra 5% fidelity. 1: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI https://github.com/RVC-Project/Retrieval-based-Voice-Convers... 2: https://github.com/w-okada/voice-changer https://github.com/w-okada/voice-changer