ccextractor/docs/BINARY_FILE_FORMAT.TXT

Starting in CCExtractor 0.52, the default binary format changes. The previous one,
whichwas ported from McPoodle's tools is too simple. It just contains the closed caption
data, with no timing. It can also contain data from one field, which means one language.

A new format, still simple but enough to transport all data with timing is as follows:

RCWT (Raw Captions With Time), version 0.001
============================================
File header (11 bytes), required:

byte(s)   value   description
0-2       CCCCED  magic number, for Closed Caption CC Extractor Data
3         CC      Creating program.  Currently Defined:
                       CC -> CCextractor.
4-5       0052    Program version number
6-7       0001    File format version
8-10      000000  Reserved

Time header, required for every group of data packets
0-7       <var>   FTS value (Time in ms)
8-9       <var>   (1-65535) Number of caption data blocks with this time
                 stamp.  If the number of blocks exceeds 65535 write
                 another time header for the remaining blocks.

Caption data
0-2       <var>   Three byte caption data.

Notes:

- Data is in process order, i.e. the 608 or 708 decoders can process it 
  sequentially. Programs generating files in this format are responsible
  for sorting the data when needed.
- FTS is the time, in milliseconds since the beginning of the file, 
  starting in 0.
- The closed caption data is always 3 bytes, the first of which is the
  cc_type and cc_valid fields and the other two are the actual data
  pairs. Depending on cc_type, they will be EIA-608 or EIA-708 data.