How is closed caption generated? Is it like the Dragon voice to text system? Or, are there happy little beavers listening to the audio and typing the text?
Both; I think.
I can see errors being backspaced out and then being re-typed during live(?) stuff - newcasts and the like.
Pre taped shows have the text generated off line; giving more time to get it right.
My Iphone does pretty well converting what I say into text during messaging.