RE: Discovery of track and named fragment names

Hi Silvia,

> -----Original Message-----
> From: [mailto:public-media-
>] On Behalf Of Silvia Pfeiffer
> Sent: woensdag 18 november 2009 23:56
> To: Davy Van Deursen
> Cc:
> Subject: Re: Discovery of track and named fragment names
> Hi all,
> It's great to see the topic of ROE come up here.
> I am in a similar situation with HTML5 and video accessibility: I
> really need to somehow specify what alternative and additional tracks
> are available both inside a file and associated external files. I am
> eyeing ROE for that, too. It's the same resource discovery problem
> that you're facing.
> Davy, how did you find working with ROE? Complicated? Or did it give
> you what you needed?

For the description of track fragments, it certainly does the job IMO.
However, I'm not quite familiar with the way named fragments are represented
in ROE (I suppose a combination of ROE and CMML will be needed?), so I
cannot say anything about named fragments (yet). 

> Why did you choose ROE and not MPEG-21 DID?

I just mentioned MPEG-21 DID as a possibility (MPEG-7 is another one I
guess), while ROE has been designed for this kind of purposes. Also, MPEG-21
DID could be used for this purpose, but then (because of its genericity) we
would have to restrict the semantics of certain MPEG-21 DID components
(e.g., define, unambiguously, how the organization of a media resource in
terms of tracks and named fragments maps to the structure of a Digital

> BTW: I don't think we need to solve it in Media Fragments, but we
> definitely need to solve it in HTML5.

As said above, different formats can be used to describe tracks and named
fragments of media resources and since this information can be considered as
media annotations, I think we should look into the direction of the Media
Annotations Working Group (also CC'ing them now). For instance, the MAWG
defines two properties that could be related to this topic: ma:fragments and
ma:namedFragments. If formats such as ROE are mapped to the Media
Annotations API, then within HTML5, they could make use of this API and do
not have to bother which format to use. For instance, by making use of the
API functions related to the property ma:fragments [2], we could then
perform the following steps to discover the track names (in pseudo-code):

DOMString[] getTrackNames(MediaResource mr) {
	Fragment[] fragments = mr.fragments();
	DOMString[] tracknames;
	for(Fragment f : fragments) {
		If(f.role = "Track") {
	return tracknames;

Does this makes sense or am I completely misusing the fragments() function
here :-)?

Best regards,



Davy Van Deursen

Ghent University - IBBT
ELIS - Multimedia Lab

Received on Friday, 20 November 2009 08:36:51 UTC