I think the problem with that is that contemporary speech synthesis is pretty limited sounding, except for the very high-end expensive products (AT&T, Natural Voices, etc)
It depends on where "expensive" is for an individual user
I use a female voice from Ivona, okay it did cost some money (49 euros iirc) but it does the job very nicely and seems good value to me.
The value proposition might be even higher for the speaker of languages that don't get direct support in-game.
Essentially, the engine is there in Windows ... it seems a shame not to use it.