[Ogg a11y] a generic mapping of text codecs into Ogg

[Silvia Pfeiffer]

On Sat, Nov 15, 2008 at 1:56 PM, Conrad Parker <conrad at metadecks.org> wrote:
> 2008/11/14 Silvia Pfeiffer <silviapfeiffer1 at gmail.com>:
>> Hi all,
>>
>> Over the last week I have developed a wiki page with a generic mapping
>> for "text codecs" into Ogg. This may include srt, sub, timed text,
>> cmml, and lots of other formats that provide time aligned text.
>>
>> The specification can be found here:
>> http://wiki.xiph.org/index.php/Text_Codecs_Ogg_Mapping .
>> I would appreciate feedback.
>
> The description of the granulepos split is not like that of other codecs:
>
> | For this, the granulepos is being segmented into two parts, where
> | one part signifies the insertion time of the current page, and the
> | other part contains the granulepos of the first page with text codec
> | data that is still active.
> |     +-----------------+-----------------+
> |     | granulepos prev | current time    |
> |     +-----------------+-----------------+
>
> Assuming you mean to use the same granuleshift scheme as CMML and
> Kate, the second half should not encode the current time, but the
> offset from the time of the previous text codec data page; ie.
>
> |     +-----------------+-----------------+
> |     | time_prev        | offset          |
> |     +-----------------+-----------------+
>
> then current_time = time_prev + offset

Ok, I have changed it. The wording is now trying to describe it the
way it is described in Kate.

BTW: if this is indeed the way it works in CMML, then
http://annodex.net/TR/draft-pfeiffer-cmml-03.html#anchor31 is wrong,
too.

Cheers,
Silvia.