This document is no longer under construction due to luck of time!!!
Hence it is seriously out of date. Mail me if this intersets you and I'll try to continue
working on it...
Please send additions to the me. Especially information on
non-audio metadata (for example examples of metadata for videos or images) is needed.
This documet should contain all links, ideas and writings about metadata for multimedia that I
find/write. This ranges from simple human-readable metadata (that's the kind we see all around) to
fully machine-processable one (probably someone already attemted implementing it but I have never
seen such a thing ;).
Also you will note a strong bias towards audio (since that's the kind of multimedia I work the
most with) and some Vorbis-realted bias (as it's the codec I use
the most). It's not intentional - if anybody can provide me with information that broadens my
horizons, I would be glad to make it more balanced.
Similar pages you should look at
Metadata Standards (not multimedia specific)
- Dublin Core Metadata Initiative - An old-time
established standard. Has a pair/value structure, with optional qualifiers on elements to
refine their exact format. Have drafts for expressing it in RDF/XML. A good
introduction to metadata in general and Dublin Core in paticular is here.
- Resource Description Framework (RDF) - a heavyweight
XML-based format that seems to be very powerful. Intended for all metadata in the world, not
just about files. Represents data as graph of relations. Lot's of work is being done on it
so we will see in the future even more.
- W3C Semantic Web Activity Statement - They
try to build a computing/networking future really full of metadata.
- ID3 - This is used with MP3.
- ID3v1/v1.1 - The original version that
prevails in the world and is the de-facto meaning of "tagging" for most people.
Sadly. It's limitied to fixed-width (30 bytes) fileds: TITLE, ARTIST, ALBUM, COMMENT,
(4 bytes) YEAR and a one byte (index into a hardwired list) GENRE field. Version 1.1
trades two bytes of the comment for a track number (almost completely backward
compatible). Goes at the end of the file.
- ID3v2 - a bloated over-engeneered
format. Has chunk container structure (unknown - ignored), tons of defined frame
(chunk) types (including synced lyrics, pictures, play counters, MPEG seek tables and
what-else), per-tag comression, ugly "unsync" scheme to prevent it from being played (since MP3
stupidly has no provisions to add metadata), goes at the begining of the file (to see
when streaming) and more... It's very binary styled (and MP3-oriented) and hard to
implement properly, which is why proper implementations are scarce (and take long to
appear). Most people still don't know it exists...
- Lyrics3 Tag v2.00 - a format for
storing lyrics in mp3. Not extremely elegant if you ask me but does the job. Note
that ID3v2 uses its own format which will probably supercede this, if players will
- Inside the distribution of id3lib you can find
somewhere a file describing the obsolete MusicMatch tag format (it switched to id3
later). One interesting feature it had were several tags defining the mood, tempo,
etc. of the file.
- Music Brainz - they use XML for metadata and have a proposal for MusicBrainz II metadata. Also
interesting are thier database
structures. They used to be called CDIndex.
- Bitzy - a generic (all file kinds whatever) database. Bitzi Metadata Resources has some good links.
- FreeDB - Seem to just freely implement CDDB (now non-free and called Gracenote), which is pretty
weak as metadata goes. CDDB2 (this is the only thing I
found about it) exists but is proprietary and seems more of buzzwords than good technology.
Ogg Vorbis mailing lists archives
I'm archiving metadata-related messages on the mailing lists since about March 2001 for my
personal use (for example to aid me write this page). It started out quite selective but soon I
understood that I need to archive everything half-related (including flame wars) to see the whole
picture. You can download it (divided by years, unix mailbox format) here.
Here I tried to list the most "enlightening" messages. Tip: explore the threads of these
messages if you have too much spare time ;-). There are surely important ones I missed.