W3C home > Mailing lists > Public > public-tt@w3.org > December 2008

Re: caption/subtitle discussion on ogg accessibility list

From: Silvia Pfeiffer <silviapfeiffer1@gmail.com>
Date: Tue, 9 Dec 2008 11:13:15 +1100
Message-ID: <2c0e02830812081613m2f2b7bc0x5c2e83152245b985@mail.gmail.com>
To: "Sean Hayes" <Sean.Hayes@microsoft.com>, "Geoff Freed" <geoff_freed@wgbh.org>
Cc: "public-tt@w3.org" <public-tt@w3.org>

Let me clarify what is happening at Ogg in more detail..

The discussions about Ogg and accessibility are motivated by the use
of Ogg Theora/Vorbis as a baseline codec in Mozilla/Firefox for HTML5
video tag support.

Mozilla is investigating how to get support for subtitles and other
types of time-aligned text (such as speech bubbles, karaoke,
hyperlinked text annotations and the like) into the Web browser.

It has been determined that there is a need for two approaches:

1) An out-of-band approach:
In HTML5, the video resource and the text resource would be linked
separately through the <video tag>. The links to an external text
resource would need to be accepted by the Web browser as a
time-aligned text format for a video and used on the fly. This can
look something like this:

<video src="http://example.com/video.ogv" controls>
 <text category="CC" lang="en" type="text/x-srt" src="caption.srt"></text>
 <text category="SUB" lang="de" type="application/ttaf+xml"
 <text category="SUB" lang="jp" type="application/smil"
 <text category="SUB" lang="fr" type="text/x-srt"

NOTE that this is a proposal, unimplemented, and not yet discussed by
HTML5. But it is an idea we are toying with at Ogg accessibility.

2) An in-band approach:
The delivery of time-aligned text would be multiplexed together with
the video file inside the Ogg stream. This will then allow the Web
browser to extract the text upon decoding. It will not change anything
in the current version of the HTML5 video tag:

<video src="http://example.com/video.ogv" controls>

For this second case, we are discussing means of including
time-aligned text (or what we call "text codecs") into the Ogg
bitstream. Which is where Geoff's concerns come in.

Currently, we have defined a generic mapping for any type of
time-aligned text into Ogg by defining OggText.
This generic mapping can in principle take DFXP or srt or CMML or kate
or SMIL or any other format. Mapping of a specific format requires
some further small specification on top of OggText.

Currently we have started with the simplest mapping, which is OggSRT.
SRT and srt-like formats (like SUB) are simple in that they are plain
text and a time segment and most media players can deal with them.
Also, a large number of available subtitles and captions online are
being provided in these formats. Also, YouTube supports them, which
will further encourage people to provide more of these.

To get a quick and effective result for Mozilla and their needs for
subtitles, srt is the most sensible choice.

This does in no way shape or form inhibit DFXP from getting supported
inside Ogg. It's just simply not first implementation priority. Also,
I am under the impresison that through the public-tt work, DFXP may
still see some changes in the near future and I am looking forward to
the final format, which will provide more powerful time-aligned text
capabilities to Web browsers. Most subtitle needs can be fulfilled
with srt, but there are other needs, which DFXP will satisfy.

Just to mention this, too: there are further needs that we have
identified, that DFXP currently cannot satisfy IIUC - such as outgoing
hyperlinks for a piece of text, or regions that when you mouse-over
make another text region appear. I may be mistaken with these though
and would be curious to find out how such requirements could be
satisfied with DFXP.

Best Regards,

On Tue, Dec 9, 2008 at 5:02 AM, Sean Hayes <Sean.Hayes@microsoft.com> wrote:
> I suspect the Ogg group will go their own way, and while it is disappointing
> they would not pick up dfxp directly I can understand their reasoning; and
> this is really not much different to any other proprietary codec.
> The primary point about DFXP for me is in its use as a clearing house
> between production and delivery, and as long as there is a dfxp<-->ogg
> translation I see no problem in them using whatever they want for end user
> delivery (although the subrip text format does seem overly basic, it's
> really not that different from 3gpp or 608.). I'd love the world to
> standardise on a single delivery format, but I'm realistic that that is not
> on the cards any time soon, it being too easy to just whip up another
> time+string format without really considering generality, users needs, IP
> protection, internationalisation etc, etc.
> The way to get to the ideal point is to start at the production and b2b end.
> The key here is a common origination format, once that is established, and
> then when mainstream proprietary players to consume and display it with full
> fidelity; then we can start to think about a one size fits all solution
> based on dfxp or some successor.
> Sean Hayes
> Media Accessibility Strategist
> Accessibility Business Unit
> Microsoft
> Office:  +44 118 909 5867,
> Mobile: +44 7875 091385
> From: public-tt-request@w3.org [mailto:public-tt-request@w3.org] On Behalf
> Of Geoff Freed
> Sent: 08 December 2008 14:09
> To: public-tt@w3.org
> Subject: caption/subtitle discussion on ogg accessibility list
> there's a lengthy discussion about captions/subtitles going on at the ogg
> accessibility list.  archives are available at
> http://lists.xiph.org/pipermail/accessibility/, or you can sign up at
> http://lists.xiph.org/mailman/listinfo/accessibility and join in.
> there has been some debate over what text-display format to support
> initially, and the group seems headed toward support of SubRip (srt).  i've
> expressed concern that doing so might initially limit the usefulness of ogg
> captions/subtitles, and have lobbied for the inclusion of dfxp from the
> beginning, rather than waiting until after srt support has been established.
>  you can see my comments in the archives.
> g.
Received on Tuesday, 9 December 2008 00:13:55 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 5 October 2017 18:24:03 UTC