Vivid dreams do make some awesome scenarios for scifi and fantasy.
Watching the trailer made me realise how comparatively hard to do this is - most pixel-art point&clicks are fine just barely animating the mouth movement, and the resulting voice (if any) doesn't even have to be synced with the mouth/lips movement at all. Compared to that, what you are doing is exponentially harder to do!
Feel free to quote me on that elsewhere.
That said (and to be fair), there are also other examples of it being hard even with classic "speaking" - the first Legend of Kyrandia had partial lip syncing in the talkie version (sometimes, it desynced). There, that fact was compounded by it running under DOS. Imagine coding lip syncing for various hardware by interrupting the CPU and writing it in Assembly or pure C. With the audio format standards being atrocious back then. :p