Comment by midoridensha

If these things were done properly by the studios, a user agent wouldn't be necessary. I've watched current TV shows with subtitles (official ones, not fan-made) and they had obvious spelling errors in them. Subtitles are a separate track, and the time they're displayed is set in the track info; showing the subtitle at just the right moment is well within their capability, but they just don't care.

But yeah, you're right: someone's probably going to eventually come up with an AI user agent that tries to do these things automatically.