What about Metallica and U2 both having a song called "one", should we add [the U2 song] and [the Metallica song] in the track title? Or when Johhny Cash covered the the U2 one, should we add that explicitly to the track title? And with live versions, should we enter the date of the performance to the track title? and if a song has been performed multiple times on one day also the approximate time?
In my opinion, this is what we have AR's for, a track title is what it is, the name of a song. We have the MB-ID as an exact identifier, and using AR's we can find out which other tracks contain the exact same music. Bram / jongetje Tim schreef: > Forgive me for beating a 2-years-dead horse, but I have not yet given > my thoughts on the issue. I believe that if there was any consensus in > the discussions I have been catching up on, it is all voices are > welcome. To begin: > > I believe that unique tracks should have unique track names across all > releases. This seems to be essential for distinguishability of tracks > by their track names, effectively describing any differences in sound > (ie unique tracks) with unique, specific track name information. (Note > I am not using the term TrackName; by "track name" I mean everything > after the artist name.) I also believe the converse: that unique track > names should be paired with unique tracks across all releases. There > should be no doubt as to what sound (unique track) one is referring to > given a track name. To summarize: every unique track should be paired > with one and only one unique track name. Each name has one sound and > each sound has one name. > > Therefore, it follows that if a physical release lists "Funky Shit > (album version)" in its tracklisting, and this recording is the exact > same as "Funky Shit" on the actual album that "album version" is > referring to, or whatever is chosen to be the default or main version, > (that is, if tracks labelled as "Funky Shit (original mix)" and "Funky > Shit (album version)" are non-unique, identical sounds), then their > names should be somehow merged conform to the above one-to-one rule > (ie, they should both be labelled "Funky Shit"; this is one pair.) > > Similarly, if two physical releases both list "Funky Shit" names but > they actually contain unique sounds, then these unique sounds should > be given unique names, overidding the tracklisting just as we would > for a spelling mistake. If one "Funky Shit" comes from a single > release and the other from a full album, then the first should be > called "Funky Shit (single edit)" (or something similar), assuming we > have chosen the sound from the album to be worthy of the default, base > name (no extra parentheticals) as we normally do. (Or, if the single > contains the default sound, then its tracklisting should say "Funky > Shit," and the album's listing should say "Funky Shit (album version)" > > > Therefore, I address all who favor full inclusion (or full removal) of > "album version" and similar ExtraTitleInformation by responding to a > list of arguments from an earlier discussion: > > "we loose version information when it's removed" -- If the track is > not actually a version of the default (ie if we do not actaully have > two unique sounds), then it should not be labelled as such. > > "in line with 'state what is on the cover'" -- Everyone seems to agree > that covers are sometimes wrong. There are misspellings and mislabellings. > > "when [album version is] removed, a release can have two tracks with > the same name, making the track listing ambiguous" -- Then fix the > mistaken listing. Either call one of the "(single edit)" or the other > "(album version)". > > "The album version isn't necessarily the main version, and the album > version may not be called an album version but instead LP version, 12" > version, etc. and in both cases the version info is kept." -- If we > match all sounds to unique names, then it doesn't matter how a track > is incorrectly labelled (LP version, 12" version, album version, or > even original mix), if it shares the same sound as the default-ly > named sound, then it should be given the default name. > > "There's currently an inconsistency in assumptions we make, i.e. an > unlabelled track on a live release is a live version, but an > unlabelled track on a single release is an album version" -- From all > of the above, it follows that "unlabelled tracks" (what I interpet to > mean default-ly named tracks, to be consistent with terminology I used > earlier) on live releases (assuming they differ in sound from the > default-ly named tracks) should be labelled (live), and unlabelled > (again, default-ly named) tracks on single releases should be labelled > (single edit) if they differ in sound from the true default; > otherwise, if the unlabelled tracks do not differ from their true > default versions, then they should retain their default names. > > I left out the middle one, as I think it's the best, and it drives to > a deeper issue: > "SameTrackRelationshipType is the AR that can state that two tracks > have the same content; no need to rename them all to the same name" -- > I wrote my first paragraph really from the "tagging" > purpose/perspective, assuming that all tracks should be uniquely > identified by fields available in ID3, and vice versa, not from > otherwise invisible musicbrainz-specific meta-data. That is, hopefully > all musicbrainz data that actually identifies a track as unique could > be contained in text as ExtraTitleInformation. If two tracks are > identical and share SameTrackRelationshipType, then their ID3 track > title field (and total musicbrainz name) should be the same. If two > tracks have been identified as non-unique with > RemasterRelastionshipType, then one of the tracks should be called > (remaster), or the other (old) or something. So I think the question > is: does the musicbrainz db extend ID3 vertically or horizontally? Can > all information that identifies a track as unique from any other be > contained in parentheticals? (If two unique tracks have unique track > names then there is no ambiguity. If one track has two names, then > there is ambiguity unless you find out someone has reported them with > SameTrackRelationshipType. Also, even if musicbrainz contains info > that will not fit in ID3 even as just lumps of text in a more generic > ID3 field, I think it would be really bad if ID3 tags explicitly > disagree with musicbrainz.) Either way, if two tracks share > SameTrackRelationshipType, then removal of "album version" etc, is not > a loss of information (as SameTrackRelationshipType would itself be > redundant if unique tracks were named uniquely), and in fact should be > done to reflect SameTrackRelationshipType (removing even partial > ambiguity), regardless of what releases they came from. The only > needed contextual information is that the one track contained in two > releases and listed with SameTrackRelationshipType (and hopefully the > same track name) are in fact listed under two different releases. > > Thank You. > ------------------------------------------------------------------------ > > _______________________________________________ > Musicbrainz-style mailing list > Musicbrainz-style@lists.musicbrainz.org > http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-style _______________________________________________ Musicbrainz-style mailing list Musicbrainz-style@lists.musicbrainz.org http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-style
