[mythtv] Ticket #10172: Bad cc708 output from mythccextractor

[Jim Stichnoth]

On Wed, Nov 23, 2011 at 2:45 PM, Daniel Kristjansson
<danielk at cuymedia.net> wrote:
> On Wed, 2011-11-23 at 13:48 -0800, Jim Stichnoth wrote:
>> Thanks.  The durations now look like what shows up during playback.
>> Interestingly, at least in this recording, the end times of the cc708
>> captions are usually within a couple frames of the cc608 end times,
>> but the cc708 start times are usually substantially later than the
>> cc608 start times, making cc708 a lot harder to use.  Mythccextractor
>> is a great tool for looking into this.
>
> Have you tried with a program that was captioned in 708? I believe
> there may be some 608->708 converter on the market which uses sets
> the start time of a 708 caption set to be when the last character
> of a 608 set was output. The 608 is a very low bitrate stream so
> you actually get a bit of time to read the first part of the caption
> before each set finishes. If the end time of the generated 708 is
> set to about the same time as the 608 so you end up having
> significantly less time to read the text.
>
> If this is the case we can adjust for that when we display 708 captions,
> but I need a bit more confidence there isn't just a bug in our 708
> decoding.

I checked a number of recordings (visually, not with mythccextractor)
and found that the 708 caption behavior depends on the
channel/network.  The short captions caused by delayed start times
happen on all recordings I checked from ABC, CW, FOX, and PBS.  NBC
has good durations, but they show up too early.  Only CBS has
consistently good 708 captions.  So your idea of a flawed captioning
device makes sense.

For making comparisons, does anyone know of a Linux command-line tool
or a Windows video player that can do 708 captions?  Windows Media
Player doesn't show any captions at all, Windows VLC only shows 608
captions, and ccextractor doesn't support 708 captions.

Jim