|
And for the last 20+ years, this is how I spend my day... authoring captions. I have transcriptionists for the typing, but everything goes thru me before it goes out the door.
Editorial choices like those noted in 12 Angry Men should not happen, of course. 25 years ago or so, captions were "edited" to deal with issues of reading speed and vocabulary because the impression was that the intended viewer needed the help. Later on, the consumers of captioning were revealed to prefer verbatim captioning, because you don't get better or faster at reading comprehension unless you're challenged by it. (There are limits, of course. And, captioning for kids' programs is a little different -- but if they're truly for kids, the vocab should be appropriate anyway.) Changing words like "get the chair" would be like colorizing movies originally shot in black and white, and what kind of idiot would do that?
The apparently dismal education level of some captioners is even more discouraging than misplaced editing or misspellings. Often they just don't know what the word is, much less how to spell it. I worked on a game show just today that had a reference to Boolean algebra. A few years ago, in a Chuck Yeager bio, he referenced a plane's horizontal stabilizer, which appeared as: "oars zontal" instead. And I'm sure that error made it through a few people. Another example: on 30 Rock several years ago, Liz Lemon ended a scene with the word "Whuck?", as a network-acceptable version of WTF. It was totally missed by the captioners, who just wrote "What?". Always important to listen closely, and there are millions of other examples.
For shows that are captioned live, there will be some mistakes due to both spelling prep, and some unusual terminology. But on a taped show, where the deadline is not "three seconds from now", it would be appropriate to do some research.
For live-captioned work, there are some cascading delays that cause the "overrun" situation mentioned above. Many live captioners work from home, and may not be listening to the program in real time. The usual human captioner delay is about 2-3 seconds, as long as the announcers aren't speaking a mile-a-minute trying to squeeze content in just before the break. So, there will be times when the broadcast delay, human delay, steno-to-english translation, and rebroadcast delays can start to add up.
As far as discerning "foreground" from "background" dialog, keep in mind that the steno captioners are bopping along at about 200 wpm or so, and deciding when one sentence ends and another begins, correctly flagging who is speaking, etc. which gets to be a full-time job. If someone off-camera starts a sentence that later turns out to be stage direction, etc., the first part of the sentence is already out the door. Although the tech does exist to "take it back", you're probably well into commercial by then, so your caption encoder is being overridden by the captions coming from the commercial, and there's little to be gained.
Keep in mind that the captions you see are being done by the lowest bidder, not the best captioners. And the sad truth is, although the law requires the vast majority of shows to be captioned, it does not mandate that they be captioned well.
And to answer the often-asked question of speech recognition captioning: know. Even though people have been selling it for years, it's not ready for prime time, and is still speaker-dependent and fails miserably if you have to stop and spell something -- which is quite often.
And now I shall dismount my soapbox.
Chip
|