This is my opinion. I personally like all the different methods you've tried and experimented. All of them with dislike and likable things about them.
My preference was the early voice acting roles with just Protag + gardevoir's original female voice (nonrobotic voice)
However, after some time, I grew to prefer no voices at all. Just text and pokemon SFX. The plain text of speeches and the expressions/body language give a lot of depth. It gives the audience the freedom to truly experience the emotions, and the story as a bare skeleton as a raw material, like reading original source material as opposed to seeing a play or musical.
There may need to be other features to add, such as 'add voice acting' or 'no voice acting' feature, and skip buttons from here or there for various scenes/chapters in each episode.
Keep up the storytelling and suspense and feels and awesomeness.
JMartin97
Nice :)