My hearing is OK (so far) but I also started leaving CC on for some scifi programs especially to get spellings of names that aren't always clear from the sound. And I found some interesting things:
Also, AIU from my reading, there is a rather low limit on the data rate for line 21, and an even lower limit on the number of lines/words per second average viewers are assumed able to read. When there is rapid complex speech it is usually simplified, sometimes sharply, or less-important prompting or confirmatory lines dropped altogether. And when there is overlapping speech (or speech with significant sound/music) pretty often only the "main" speech is captioned.
VHS definitely, and tapes I record from "air" include it along with the rest of the video. I don't think DVD actually has the syncs and (thus) BIs recorded; I believe caption is a separate data stream that is synthesized into line 21 by the player. I see _very_ low garbles on DVD play, maybe .01/hour, always corrected on replay and hence likely faults in my TV (monitor) compared to usually 1-10/hour for "broadcast" (analog cable) some but far from all of which I find recur on rebroadcast in cases I can and did check -- presumably this is either recorded badly on the master or there are data or video patterns or both especially prone to (self) interference or noise.
Older material like old movies must have captioning added. For programs created in the last few years, after the FCC mandates for it became effective, I believe it is mostly done during the final stages of production. In some cases I see captions that are quite different from -- and not just shortened forms of -- the spoken lines, which I suspect were ADRed late but the captioning not changed; and since caption data must be transmitted a second or two in advance in order to appear en-bloc in sync with the video, sometimes I see at the end of a scene a line obviously for a following scene that has been cut. Conversely this is why a line immediately after a commercial break usually isn't captioned.
ObSemiTopic: According to my reading much of the realtime captioning is done by telecommuting. And I have indeed a few times seen Hayesisms (NO CARRIER, CONNECT nnn) in the captions.
- David.Thompson1 at worldnet.att.net