>>6237823Yeah, I've been trying to gather informations, but it seems it's speech to text then text to speech, all in one, while including the text in the overlay. It's pretty interesting. It works pretty well most of the time too. It could be used with machine translation as well.
