They won’t because of hallucinations. They could work in mature games though where its expected that whatever the AI says is not going to break your brain.
But yeah a kid walks up to toad in the next Mario game and toad tells Mario to go slap peaches ass, that game would get pulled really quick.
I just re-read my comment and realised I was not clear enough.
You bundle the text and the AI-TTS. Not the AI text generator.
The content is… AI assisted (maybe a better way to put it).
And yes, now you don’t need to get the VA every time you add a line, as long as the License for the TTS data holds.
You still want to be having proper VAs for lead roles though. Or you might end up with empty feeling dialogues. Even though AI tends to put inflections and all, from what I have seen, it’s not good enough to reproduce proper acting.
Of course that would mean that those who cannot do the higher quality acting [1] will be stuck with only making the TTS files, instead of getting lead roles.
But that will mean that now, places where games could not afford to add voice, they now can. Specially useful for cases where someone is doing a one dev project.
Even better if there can be an open standard format for AI training compatible TTS data. That way, a VA can just pay a one time fee to a tech, to create that file, then own said file and licence it whichever way they like.
e.g. most Anime English dubs. I have seen a few exceptions, but they are few enough to call exceptions ↩︎
Oh come on, LLMs don’t hallucinate 24/7. For that, you have to ask a chatbot to say something it wasn’t properly trained for. But generating simple texts for background chatter? That’s safe and easy. The real issue is the amount of resources required by modern LLMs. But technologies tend to become better with time.
