ogg dev - Jan 2008 - Ogg/Kate preliminary documentation

Hi,

I've now uploaded the preliminary documentation on the xiph wiki:
http://wiki.xiph.org/index.php/OggKate

Attached is the current source tree for the libkate library.

The tarball also contains the patch to oggmerge (which you will need
to apply if you want to merge Kate streams with Vorbis or Theora streams)
and the patch to MPlayer to use Kate streams as subtitles.

An example is included - note that the text format used to define the
stream is ad-hoc and the encoder/decoder code very rough and
temporary. No error checking in those tools, the input is expected to
be error free.

I look forward to your comments.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kate-preview-noexample.tar.gz
Type: application/x-gzip
Size: 28024 bytes
Desc: not available
Url :
http://lists.xiph.org/pipermail/ogg-dev/attachments/20080115/592477fd/kate-preview-noexample.tar-0001.bin

Hey,

On 1/15/08, ogg.k.ogg.k@googlemail.com <ogg.k.ogg.k@googlemail.com>
wrote:
> I've now uploaded the preliminary documentation on the xiph wiki:
> http://wiki.xiph.org/index.php/OggKate
Really nice work, by the way.
> The tarball also contains the patch to oggmerge (which you will need
> to apply if you want to merge Kate streams with Vorbis or Theora streams)
I have looked into the patch.  It doesn't take into consideration
neither Skeleton, which is used now in pretty much everything encoded
in Ogg (except for single stream Vorbis and Speex files), nor the file
extension for Theora, which is now .ogv.

To be fair, it's much better than the current code, but you may want
to extend/modify the patch taking these into account.

A bit of background information: Skeleton is an extension of Ogg to go
around certain limitations of the container.  It helps the detection
of the content within, as well as gives the ability to choose which
stream to play (imagine two audio tracks for two different languages
akin to DVD Video).

.ogv is the file extension for anything video in Ogg.
> I look forward to your comments.
I have to study more extensively the documentation.

Meanwhile, here's some comments:

* don't bother with non-UTF8

* I agree that CMML is complex for something as simple as karaoke, but
I am of the opinion that the more formats doing the same thing, the
worse it will be for their adoption rate.  Developers just won't
bother to support everything out there.  Some may support CMML, some
may support Kate, some may support none.  Then we have the users not
knowing which format to choose for lyrics/subtitles.  I think that
perhaps it would be better to join efforts on an unique project.

-Ivo

On 16/01/2008, Ivo Emanuel Gon?alves <justivo@gmail.com>
wrote:
>
> * I agree that CMML is complex for something as simple as karaoke, but
I disagree that karaoke is simple; to do karaoke properly is about as
complex as text codecs get.

I also disagree that CMML is complex, but I may be mistaken. In terms
of complexity, I see it as about 10% of the way between using
zero-markup and using HTML.

cheers,

Conrad.

Thanks for the feedback,
> I have looked into the patch.  It doesn't take into consideration
> neither Skeleton, which is used now in pretty much everything encoded
> in Ogg (except for single stream Vorbis and Speex files), nor the file
> extension for Theora, which is now .ogv.
To be honest, I just added Theora because I needed a simple way to
multiplex streams. Also, it'd likely take a rewrite of oggmerge to make it
usable for multiplexed streams, I think it assumes a single logical stream
per physical stream, though I didn't have an in-depth look at this issue.

I did see references to Skeleton, I'll have a look at it. I didn't
realize it was
used widely - I've certainly not seen anything with it in it (though I
don't
tend to download many Ogg streams, I encode test ones myself :)).

As for .ogv for Theora, do you mean Theora-in-ogg, or raw Theora ?
If Theora-in-ogg, then what about Theora multiplexed with other stream
types, like Vorbis, since a lot of videos will have sound embedded ?
> * don't bother with non-UTF8
Yes, I was going to leave it as is while keeping the header field there,
just in case. I do know, however, that people in eastern countries tend
to dislike utf-8 for the size it takes for their language, as it's been,
er...
/optimized/ for the latin alphabet. I am also not clear if every code point
can be coded in utf-8, I'll have to dig that up.
> * I agree that CMML is complex for something as simple as karaoke, but
> I am of the opinion that the more formats doing the same thing, the
> worse it will be for their adoption rate.  Developers just won't
Yes, this is a fair point, but then you have Vorbis, Speex, and FLAC,
each with its own particularities. Yes, Vorbis would probably yield bad
sound at the low bitrates Speex works at, etc, this is just an example.
> bother to support everything out there.  Some may support CMML, some
> may support Kate, some may support none.  Then we have the users not
> knowing which format to choose for lyrics/subtitles.  I think that
> perhaps it would be better to join efforts on an unique project.
I understand where you're coming from.

Thank you for your comments.

> I have looked into the patch.  It doesn't take into consideration
> neither Skeleton, which is used now in pretty much everything encoded
> in Ogg (except for single stream Vorbis and Speex files), nor the file
> extension for Theora, which is now .ogv.
I've added the [xgav] variants, the attached patch replaces oggmerge.diff.
I haven't looked at Skeleton yet though.
FWIW, when I have split streams, I usually name them foo.vorbis.ogg,
foo.theora.ogg, etc.
Maybe the patch should try the magic check for all files ?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: oggmerge.diff-with-ogv
Type: application/octet-stream
Size: 2209 bytes
Desc: not available
Url :
http://lists.xiph.org/pipermail/ogg-dev/attachments/20080117/356737ea/oggmerge.obj