Where can I find information (in English or at least translatable to English) on what is involved in training new voice databases?
In a vague sense, Synthesizer V AI voice databases are created by using machine learning to analyze vocal sequences. Unlike older concatenation-based singing synthesizers (such as Vocaloid prior to version 6, and the original SynthV “Standard” voice databases), the recorded samples for AI voice databases are actual singing, rather than individual phonemes in isolation.
Beyond this, there is no public information about the specifics of voice database development, and all third party developers that work with Dreamtonics are under strict NDAs.
The dev tools are not available to the public and the technology is strictly proprietary.
1 Like
Thanks for the information. I was hoping to be able to train up new databases through AI/ML, but it sounds like that option is not open to me.
I bought the pro version and have two voice databases that I have purchased, but there are maybe 3 or 4 more out of all of them–both from Dreamtronics and third party–that I would consider purchasing at this point.