Originally Posted By: Mark Hayes
At first, it sounds like you are chatting with a robot. But as the training progresses, the singer learns from you and emulates you.

There's already "voice cloning" technology, and it only takes less than a minute of training data to create a credible clone of a voice.

Check this list out for examples.

That's not the same, but it shows that the technology is feasible.

In a similar vein, the developers at SynthesizerV just announced they have implemented cross-lingual synthesis. That means that voices that were created in one languages can now sing in another language:



There are already voicebanks which already have "English" versions by singers who aren't native English speakers, but they often have heavy accents, and poor pronunciations of phonemes like /r/. There are still slight accents in the cross-lingual versions, but they much more usable.

-- David Cuny
My virtual singer development blog

Vocal control, you say. Never heard of it. Is that some kind of ProTools thing?