Doubtful.

A handful of new 'AI' voices are much like procedural generation, but when you have enough of it, you see it all feels the same and may even grind on your nerves; No doubt they have their uses, but it shouldn't be the go-to thing. Say dead voice actors who couldn't finish the last of some series or game, cameos or something. But it shouldn't be the default.

Vocaloid also comes to mind. Once voice which can probably do anything done after a specific person, which has spawned tons of music under a '16 year old pop star' avatar, but ultimately all sounds the same. (hence i think they have like 4 voices now to choose from)

You'd probably put a lot of care in how it sounds, you may get flat performances or ones that don't fit, which may be MORE work than hiring an actual voice actor or having someone do it.

Hmmm i listen to a lot of D&D stories, and a lot of channels did Text To Speech, and while they sounded the same it also didn't sound natural. Later they started reading out and doing all their own content instead of the more lazy result after they were established. Yet curiously this is where this thing would likely more shine, reading books/content aloud for you as a TTS replacement rather than as a product for video games. They can also be placeholders until someone can go in and actually replace it with proper acting. Or for some low budget VN's, they probably would work fine... maybe...

No, these 'cost cutting measures' are not going to cut costs. Ultimately i bet unless you are an indie group who don't have the budget for a bunch of voice people and have exhausted all your own options, this probably won't be that useful, as most cost cutting measures tend to lead to more cost cutting, which ends up destroying the very thing that made the company and products unique and they die. Feels like I've seen this a thousand times.

Anyways, gave this thing a try, reminds me of the Jordan Peterson neuron net voice AI (which is disabled right now) which after that came out a few youtubers had Jordan as a 'guest' which was obviously synthesized but was hard to tell so unless you kept up with what was going on.

But i gave it a few text feeds, and they sound... competent. Good enough to read to me, not enough to rely on for something needing more emotion or individuality. I tried two main texts on a few different 'voices' and some were hilarious and flat pan 'read off the page', while others sounded bored or too fast and dynamic.