Jason A. Hockman MUMT 611 Presentation 1 Community-Based Discography

							                                Jason A. Hockman
                                    MUMT 611
            Presentation 1: Community-Based Discography Databases
                             Professor Ichiro Fujinaga
                                 10 January 2008


Introduction

  • this presentation will introduce websites which specialize in the sharing of metadata
  to users of an online community

  • def: metadata is data generated to explain data and differ from media to media. for
  music, artist, song and album names serve as metadata which is used to effectively
  organize databases of all sizes

  • the sites discussed will vary in style and information provided, which I see to be
  related to the intent of the site


MusicBrainz

  • developed by Robert Kaye, as his work for CDDB meta tagging (now Gracenote)
  became increasingly commercialized, and out of the hands of the consumer

  • funded by the MetaBrainz Foundaton non-profit organization

  • Kaye says that it is a direct outgrowth of the cratedigger, or collector philosophy, and
  the need for accurate information about releases

  • deals exclusively with metadata, and contains no audio - differentiating it from
  several other sites which use this type of service as a front end to a store

  • created as the open-source competitor to Gracenote

  • allows users to search metadata for any CD/Vinyl/Digital Media release through
  various search criteria

  • if this information does not exist, it is quite easy for registered users to enter the
  information in community based environment similar to wikipedia

  • once user enters information, it undergoes a process of verification by other users,
  who can vote on the new information over a period of 9 days

  • the database is also available for download and site mirroring so long as its not for
  commercial interest

  • currently last fm is hosting the only mirror, and MB is looking for an east coast mirror
Picard Tagger

  • for tagging purposes, there is Picard Tagger, dubbed as the "the next generation
  musicbrainz tagger"

  • allows user to retrieve and edit/ replace metadata values, through information on the
  musicbrainz server

  • Picard does not perform fingerprinting - it uses PUIDs created by MusicIP

  • PUIDs: 128 bit data generated via MIR analysis, and is NOT audio fingerprinting

  • Picard is straightforward, cross-platform tagger, supplying PUIDs (Portable User
  IDs) to a variety of formats including: .flac, .mp3, .mp4, .ogg, .m4a, .m4p, .wav, .wma

  (demonstrate how to use picard and MusicBrainz)

** googlefight **


Discogs

   • founded and developed by Kevin Lewandowski

   • commercial site, where musicbrainz actually retrieves much of its information about
   tracknames, artists, and releases

   • while MB was designed to provide metadata to a tagger such as Picard, discogs
   was designed simply to provide discographies for CD/Vinyl/Digital Media releases

   • thus discogs has a more collector-based approach

   • has since seen growth due to web presence, a marketplace, and advertisements

   • began as a rock and dj music service, because of the overwhelming number of
   singles and a need for organizing catalogue numbers from the labels

   • has since been extended to jazz and classical

   • more detailed credit roles and ability to add additional text to define role

   • host several images

   • can use multiple artists for the same track

   • the bad: use artist names for page urls -> terrible for non-latin characters

   • discogs also has an API as well, which retrieves xml files from specific HTTP pages
      and parses them for specific metadata
Statistics

   • http://musicbrainz.org/stats.html

   • http://www.discogs.com/


Other Music Database Sites

  • mystrands: metatagging app that allows users to chat/listen on a live social network

  • allmusic: front end for AMG technologies (tapestry collection playlist generator, lasso
  music recognition service)

  • rolldabeats
            • began at the same time as discogs, under the name: tarzan
            • genre-specific database for drum&bass/jungle
            • community provides additions or corrections through use of a forum


Conclusions/Notes

  • if community membership and growth is an indicator of the appeal of these sites,
  then there really is no end in sight for their viability and scope

  • in June of last year, MusicBrainz saw a partnership with LastFM, and BBC, which
  brought MusicBrainz into self-sustainablity, no longer holding fundraisers