Talk:Cover Relationship Type

This page has not been reviewed by our documentation team (more info).

Link to which version?

There were three suggestions for StyleGuidelines for this relationship. The first was to only ever link to the very first released version. This is the policy used by the CoversProject.

The second possibility was to link each cover to those versions that have most influenced it - so most artists were influenced by Marvin Gaye's version of "I Heard it Through the Grapevine" rather than Gladys Knight's version, and so they should relate to the former not the latter.

(There was a third, somewhat radical proposal namely: covers that retain the same name as the original should be linked to the original artist e.g. Strawberry Fields by Candy Flip is a cover of a song by The Beatles)

Original suggestion:

"If you want to know who composed the track, you chase up the original version and find out who wrote that. It is generally inappropriate to give any kind of Composition Relationship Class credit except "arranged by" for a cover version. You certainly shouldn't give a "composed by" credit to the original composer, because that fact is already in the database, and duplication of data should be avoided if possible."

Original Suggestion:

"When there are several different versions of the same song, every version should be linked only to the very first released version. It is true that some covers are actually covers of other covers, but this is a subjective thing, whereas deciding which is the first released version is much more clear cut, and easier to verify independently. For the same reason, we use the date of release, not the date when the song was first written or recorded. Again, release dates are far easier to objectively verify (they're generally printed on the album), as opposed to writing or recording dates."

What do we do when a famous band covers a traditional song? I'm thinking of Thin Lizzy's version of "Whiskey in the Jar" here. I'm sure that there are more traditionalist recordings floating around somewhere, but identifying which was the first released would be a real trial. This also wouldn't make much sense, since every version ever recorded is really a cover.

We should in any case distinguish between a cover version and a parody. For example, Enter Madman 1 & Enter Bluesman 2 (by Excrementory Grindfuckers) has exactly the same lyrics and almost the same meldoy as Enter Sandman by Metallica, but the way it is performed clearly makes it a parody. Therefore adding an AR saying it were a cover version would create a wrong impression of the song. --derGraph

The CoversProject site seems a little moribund. I recently found Second Hand Songs who seem to have continued the idea. Although they only have indirect submissions through their forum. --Chiark

From Wikipedia: "In pop music a cover version is a new rendition of a previously recorded song".

In MusicBrainz covers are recorded with an AdvancedRelationship using the CoverRelationshipType. The question that this page deals with is how we should define a cover. There seem to be two possibilities:

The system for linking cover versions proposed on AdvancedRelationships seems to cover only a subset of use cases and calls for subjective decisions that complicate the moderation process. Some kind of wider, less specific and direction-less link between all versions of a song would seem more appropriate.

The two key problems are identifying the "original" track and the subtly different problem of deciding who covers who.

Some odd cases:

Using the rules from the CoversProject cases like these can be decided objectively: Their simple rule is that the first person to release a song is considered the "original". Case in point: apparently Danzig wrote a song ("Thirteen") which Johnny Cash released in 1994. In 1999, Danzig also recorded the song, so technically (for the purpose of The CoversProject) they covered a Johnny Cash song that happened to have been written by themselves.

The problem with this (and a few of the other problem cases listed above) is, I think, that "to cover" a song means slightly more than do "a version of" or it does to me at least. I think it suggests some kind of link between the band and the named version, rather than song A is the first released version of song B. (Note that if the first version released is considered important, which I think that it is to a certain degree, then this can/should be calculated by the Database from release date info, rather than being redundantly stored as part of the relationship and creating possible data inconsistancies)

This discussion stems from before AdvancedRelationships were released

Reading the above, I would suggest instead distinguishing "recordings" from "compositions". All cover recordings of a work would then be related to the same composition, as would "the original." --JoeG

Thinking of Jazz, that seems very logical to me. In Jazz musicans don't do covers, they make versions of songs but they give credit to the original composer. So I would argue that the ComposerRelationshipType is factual, and that the CoverRelationshipType includes the opinion of the moderator, which song was the original for that specific recording. --DonRedman


The question of parodies was raised on the UsersMailingList. These possibly count as covers too, although there's no consensus yet. --MatthewExon


I think we also need to distinguish, based on the initial definition, that a cover is not an artist recording their own song again. "A new rendition of a previously recorded song" could be mis-interpreted that way. -- BrianFreud

There is also a difficulty determining what the "original" recording would be for interactions between classical and non-classical recordings. When a rock band covers a track from an opera, especially if that opera was composed prior to the age of music, there would be no target recording that could be pointed to under the current definition. -- BrianFreud

This page is transcluded from Talk:Cover Relationship Type.