Ace Studio probably just added an RVC option ("voice changer") to train your own voice, so they probably just use one of their voices then run it through RVC. Plenty of high quality RVC options available, including open source that can be run locally, if you have a good Nvidia computer and you're willing to wait.That's really what I meant. ACE Studio do not really disclose what transformer model they utilize. But they allow you to train your own voice, so I am assuming it is based on newer technology.I'd say the newer technologies that might eventually threaten Synthesizer V and Ace Studio are the LLM based generators, since they seem to adapt to the emotional nuances statistically associated with lyrics, once they're able to let you consistently determine and edit the vocal melody, tempo, etc.
But the more interesting solutions will come from companies like ElevenLabs, for example. It's really just a matter of time until that happens.
The company that makes Synthesizer V is beta testing a new, more flexible version of RVC which they're calling Vocoflex. Unlike standard RVC you can edit the parameters and morph between vocals. However the current version operates in "real time" at expense of audio quality I think. Hopefully they'll add a higher quality option for rendering.
https://www.youtube.com/watch?v=ORiIsU2D0Sk
Statistics: Posted by Ou_Tis — Sat Jun 29, 2024 1:15 pm