Earlier quoted context omitted.
There aren't super technical burdens to speech synthesis in games; for example, Jackbox Party Pack 7 runs just fine on the Switch and contains a speech synthesis engine for "Blather Round". I imagine it's mostly a licensing/cost thing (since a "voice" for a speech synthesis engine still requires hiring an actor and doing a recording session).
> since a "voice" for a speech synthesis engine still requires hiring an actor That's what I'm talking about. Surely we can do away with that if we try, just as we don't need real people to build 3D models -from- if we don't want them.
Just like how we invented an “automatic programming” system where the computer will do programming for you, and it turns out that once we’ve made automated programming we have more programmers and not fewer. The tools for making 3D models are getting better and easier to use, and as photogrammetry is being used more and more, we see larger teams of modelers, not smaller.