AI voice has come a long way. It’s very difficult/time-consuming to train still, but sites like 15.ai prove what is possible with a lot of time and effort. Even that site’s best voices - the ponies - still sound robotic, but if you intend to generate masked Kylo dialogue that wouldn’t be a problem. I believe it will be feasible within a few years, as I imagine more consumer friendly tools for this sort of thing are being developed as we speak.
That’s interesting to consider, and a fascinating process. I’m hoping it won’t have to come to that - I have the lines I need the trick is making them sound like they belong together (and they’re pretty close already). My issue is the fine tuning.