Tangentially related, I was always intrigued whether it is possible to “record” someone’s inner monologue, when they actively move their vocal cords/folds, but not produce any sound. I think that you might get enough data to reconstruct sentences given that know which language to find words in.