[vorbis] xml stream formats

Ralph Giles giles at snow.ashlu.bc.ca
Mon Jul 10 15:57:28 PDT 2000



On Mon, 10 Jul 2000 robert at moon.eorbit.net wrote:

> > Having taken a look at the spec, I'm really liking RDF, both as an
> > interface to the online disc metadata database, and as a format for
> > metadata *and* stream identification in Ogg. [etc.]
> 
> I would agree with you there. However, the issue of how to handle
> time-coded data is still up in the air. Do you have any feelings on
> that? Personally I feel that RDF is too heavy-handed for dealing with
> the time coded data. RDF chunks? XML chunks? Something completely
> different?

No, I don't think we should use RDF for the time coded data. The two
proposals I see as viable are: 

Generalized xml documents which include timecode tags. This is what I've
been arguing for.

A "chunk" format with external timecode information. The chunks may or may
not include xml or another kind of markup. (I'd go with plain utf-8 text
here, myself.)

We don't have a consensus on which of the two is better.

> If we decide to go with RDF, I think it makes sense to define an RDF
> 'vocabulary' that acts as the guideline for encoding the common pieces
> of metadata. Even though RDF allows the user to make up the properies at
> will, this probably isn't desired. The players will then have to deal
> with the age old issue of 'what of this information do I really need?'.
> If there is only one way to encode the name of the artist then the
> players will know what data chunks to look for, no?

Absolutely. We need something like the Dublin Core (http://purl.org/DC/)
for media metadata. I'd suggest that we stay name-compatible with dc where
we have overlap and just add what's missing.

> > In response to the "we won't get an xml parser into winamp" argument, I'd
> > suggest that we don't have to. A simple, audio-only ogg player could just
> > ignore the metadata stream; the info in the Vorbis comment header provides 
> 
> That's true -- however, do we really need to get an XML parser into
> WinAmp? Wouldn't it suffice to get a small parser like expat into
> ogg/vorbis plugin? Let the plugin deal with the XML/RDF parsing and then
> hand the metadata to WinAmp in whatever format it wants it in.

How small can expat be? Michael's argument is that Winamp won't take a
player plugin over 70-100 KB into the main distribution, so if we could
get one in 20K we might have a chance. My interest (see above) in getting
in a general parser had as much to do with the abitrary associated xml
streams proposal above as metadata in particular.

I've also been assured that by making some reasonable limitations and
assuming a particular dtd/schema we can write a parser for our RDF
metadata in one or two thousand lines of C. I don't think there would be a
significant space issue in that case.

> > I would urge you to design your database to handle music and movie 
> > metadata on equal footing in the next incarnation. AFAIK, the DVD format 
> > encodes *no* textual metadata at all, so there's just as much of a need
> > for that as for the CD database.
> 
> I see your point -- even if it is for completeness sake. The imdb
> already does a good job of covering the metadata for the movies, so
> another database that does the same time doesn't quite seem warranted.

Ah, I didn't realize the imdb made its data available (well, I haven't
read the license, but at least you can download a snapshot). They still
don't seem to provide a machinable query interface, and afaik don't track
individual releases (on vhs, dvd, film, etc.) so I still think there's a
need for this, even if you got your primary data from them.

I guess I should write them about it?

 -ralph

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.



More information about the Vorbis mailing list