Came across this article in my news feed. It's about how Amazon has officially released a component for their game engine Lumberyard that utilizes text-to-speech for a kinda automated voice acting. Text-to-speech is hardly new and one can hear it is computer generated (transition between words is still unnatural) and the article itself mentions that there are better systems out there, including Alexa. Though the ability to adjust individual single words lends it in the long run to become better once they get the actual translation to sound fully natural, of course keeping the more machine sounding one as an option should still be there so that we can have "proper sounding" speech for machines.
The bigger question here though, imo, is when Unreal Engine will get and how soon after that MWM will start to incorporate it. I mean, if they only need to set a speech profile per character and then give them a bunch of scripts (potentially being the actual quest/dialog text we read) then I see no real reason to not have voice overs. How long it will take MWM to actually start on this after release will most likely depend on how many voice profiles there is by default and how much they have to do by themselves.
Though in all honesty I think these kinds of systems are a couple of years away still until they become fully capable of replacing voice actors. I biggest hurdle I believe is getting the proper emotional feel in speech and that will be hard for a machine.