Wasn’t sure where to put this but I found this thread and it feels right. I fell into a youtube rabbit hole tonight centered around the Japanese Vocaloid community. The software for speech synthesis has progressed farther there than I had dreamed possible. In particular I found this software, Synthesizer V Saki AI which if I understand correctly uses machine learning to tune the speech engine.
The results are astonishing, but it’s pretty hard to glean much technical information since I cannot read Japanese. This seems to be a relatively recent development as most Youtube demo’s are from this previous two weeks. If anyone has more accurate information I’d be very curious to know more!
Sorry to necro this thread but @lettersonsounds your demo at the top sounds incredible!