Music: Sraping vs Embedded Metadata Tags
#1
This post was requested due to discussion of a trac ticket 14708 against XBMC 13 (Gotham).

Currently in XBMC 12 (Frodo) as I see it, music media files are scanned into the library using data extracted from embedded metadata tags such as ID3v2.3 in mp3 format files. The metadata is used to populate the key fields in the database, such as album, song title, track, artist, and album artist. When album artist is not provided it is inferred from the artist field of files with the same album field.

XBMC 12 then runs 2 scrapers, one for "artist" and one for "album". The purpose is to populate additional data fields in the database and to locate thumbnail artwork. Thus, the "artist" and "album" data are not specific to any one music file. This is a bit different from how scraping works in video database. For the most part it seems like the "universal" scrapers are preferred. Those scrapers are "universal" in the sense that they will query various sites while scraping (but not last.fm any more for some reason even though the addon.xml says it does).

In my testing of XBMC 13 (Gotham), it appears there has been a change of functionality. The goal of the scrapers seems to be to obtain artist and album "MusicBrainz ID" from the MusicBrainz site and then use data from MusicBrainz to override some if not all metadata found in the tags. Apparently this is intended not just as a process done during initial file scraping, but would be "dynamic" in that any changes from MusicBrainz would be forced into the local XBMC music database. At present there is no option for a user to avoid implementing this change in behavior.

I can see for some users, this change would be considered a convenience. This would reduce the need to use separate tagging software (assuming the user wants to use MusicBrainz data in his/her tags). I note that existing tagging software supports natively or via plug-ins obtaining tag field data from MusicBrainz.

Of course, this assumes (or at least would be an advantage) that MusicBrainz data is canonical. I don't think this is the case. There are a number of cases (at least for me) that MusicBrainz standards (what they call style guide) results in data that doesn't work for me. For this reason, when I tag music files, I make a decision whether or not to accept MusicBrainz data into my tags (FWIW, I use a combination of software for various reasons, but for accessing MusicBrainz I use mp3tag). I have carefully tagged my music files to optimize file management and playing. Having XBMC replace my tags unilaterally with MusicBrainz data results in difficulty finding artists and albums within XBMC. Assuming I still want the additional data that scraping provides (discography, bios, art, etc) it is difficult to see if it is even possible to write or fork a scraper that would avoid modifying the tag data within the music database since it appears that use of the MBID is going to be tightly coupled into maintenance of the music library in XBMC.

scott s.
.
maintainer of skin  Aeon MQ5 mods for post-Gotham Kodi releases:
Matrix see: Aeon MQ5 Mod Matrix release thread
Nexus see: Aeon MQ5 Mod Nexus release thread
Aeon MQ 5 skin and addon repo 11.1.0
Reply
#2
This behaviour was recently changed so musicbrainz is only overwritten if we store the tags I think.
Reply

Logout Mark Read Team Forum Stats Members Help
Music: Sraping vs Embedded Metadata Tags0