Free the mouse Replay ClosedCaption
Home | Changes | Index | Search | Go
It is currently unknown exactly how the Replay stores Closed Caption data.

Since the vertical blank (VBI) isn't (and shouldn't be) encoded as part the MPEG-2 video stream and there isn't a third data stream in the MPEG file the location of this data is currently unknown.

The other associated file, the NDX, does not have the space to store the CC information either.

Technical Data:

From http://www.robson.org/capfaq/technical.html

What is this "line 21," anyway? The vertical blanking interval (VBI) of a television signal is the part "between frames." If you have an older television with a vertical hold adjustment, and you tweak it until a black bar rolls across the screen, that black bar is the VBI.

The VBI consists of a number of "lines" of video. The 21st line has been allocated to closed-caption information. The method of encoding used in North America allows for two characters of information to be placed in each frame of video, and there are 30 frames in a second. This corresponds (roughly) to 60 characters per second, or about 600 words per minute. That sounds like a lot, until you realize that it also takes one frame to transmit a command (like "go to a new line of roll-up"), and position information on pop-on captions take more than one frame. Also, that 600 words per minute includes everything required for caption 1, caption 2, text 1, and text 2, because they're all stored in the same place, so there are decided limitations on the capacity of Line 21 for captions.

Caption 3-4 and text 3-4 are stored in field 2/line 21, a concept which is probably beyond the scope of this FAQ.


What are field 1 and field 2? Each video frame has two fields. Caption 1 & 2 and text 1 & 2 are stored in line 21 of the first field in each frame. The second field's line 21 can be used for caption 3 & 4 and text 3 & 4.

Each field actually has an effective throughput of roughly 600 words per minute, so it is theoretically possible to achieve four simultaneous caption channels at speeds approaching 300 words per minute each. In reality, the commands and intermingled text channel information, along with XDS (extended data services) data cuts this significantly.


What characters are valid in Line 21 captioning? The character set was designed for the United States, and really has very little beyond basic letters, numbers, and symbols. There are some accented letters (lower-case vowels with acute and circumflex accents, a and e with grave accents, c-cedilla, and enya), but not many. See the Line 21 Captioning Character Set for the full listing.

The new EIA-708 standard for DTV captioning has a dramatically enhanced character set. Of course, you'll have to get one of the new digital TVs and watch DTV programming to get this capability.

http://www.robson.org/capfaq/caption-charset.html


-- LeeThompson - 02 Sep 2002


The closed captioning is embedded into each picture in a gop, it falls under the user_data field. The format is different between the 4k and 5k mpeg streams.

I'm still unsure how the 3k encoded CC data, I doubt it's extracted from the encoded file.

-JeffD

-- JeffD - 24 Jul 2003


Your post will appear before this form in chronological order (newest at bottom)

Topic ClosedCaption . { Edit | Attach | Ref-By | Printable | Diffs | r1.4 | > | r1.3 | > | r1.2 | More }
Revision r1.4 - 25 Jul 2003 - 03:31 GMT - JeffD

Copyright © 2001 by the contributing authors. All material on this collaboration tool is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback.