Metadata format (was Re: [vorbis] Request for Standardization: classical music TAGS)

volsung at asu.edu volsung at asu.edu
Tue Oct 2 21:42:16 PDT 2001



On Wed, 3 Oct 2001, John Morton wrote:

> > And we haven't even
> > *touched* lyrics yet.  (By the way, limitations in vorbiscomment prevent
> > you from putting the lyrics into a single tag.  Yes, I tried it once,
> > though not on Tori Amos.  And you can't spread the lyrics across multiple
> > tags of the same name, because the tags come back in a random order!)
> 
> I don't think there's any particular reason why you couldn't, assuming the 
> lyrics tag fits inside the (ample) space limitation. This could be a library 
> bug, or a problem with vorbiscomment.

The library should permit 2^32-1 bytes for any particular comment.  However,
you _SHOULD NOT_ encode lyrics using vorbis as that is very much not what the
comments are for.

It looks like the flexibility of Vorbis comments is seducing people into
wanting to use them improperly.  Many of things being discussed here should
really go into the metadata stream that people have discussed.  However, since
there is no metadata code available, we can't point people at some program and
say "That's what you should use to record X about your music."

Given that, I would suggest we refocus part of this discussion onto the
metadata stream, which needs to be agreed upon.  I've not been paying close
attention to the metadata discussion, but let me see if I can summarize the
various points I can extract from the fuzzy list archive in my head:

* Metadata about a stream should be stored in a separate logical bitstream and
NOT be shoe-horned into the comment headers.

* The metadata stream should be some sort of text format, and probably be
UTF-8.

* It has been suggested that the format be some sort of XML format.

* As a starting point (and perhaps a finishing point), Ralph has suggested
looking at the musicbrainz format.  I would _strongly_ suggest we adopt it if
it looks adequate simply because then we have a free source of data.

So what remains to be done:

1. Select a metadata format.

2. Decide upon an appropriate way to packetize the metadata stream for
inclusion into an Ogg file.  This includes answering the questions "How do I
break up a metadata stream into packets?" and "Where do I put the packets in
the Ogg stream?"

3. Write a tool to take a text file containing metadata and insert it into an
Ogg stream.

4. [Optional] Have a party creating all sorts of neat tools to automagically
grab metadata from musicbrainz and stick it into Oggs.  (Imagine if abcde
could go onto musicbrainz.org and automatically grab metadata for your files.
Or scripts that would create webpages out of your music collection.  Drool..)

So to begin with: 

Is musicbrainz a good metadata format to work with?


---
Stan Seibert

--- >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