This proposal was created to try and address some of the weaknesses in the current system:
- handling of albums being released under multiple names.
- handling of box sets, in particular when the box set combined already released albums.
- a release with two or more albums is presently treated as two seperate releases when they are really part of the same release.
- identification of re-releases and re-mastered versions as a seperate entity to the original release.
- the release attributes (Album, Single, Remix, Bootleg, etc) being unable to cope with certain valid combinations, and missing a number of useful possibilities which are usually bled into the album title.
- no support for record label information.
- no implicit support for non-CD based releases (ie: LP records, Internet releases etc are not well supported).
- restrictive country of release metedata.
This proposal has also been discussed in an extensive email thread.
- 1 Some definitions that will be used in this proposal.
- 2 Data Model
- 3 Details of a Data-Set
- 4 Problems to resolve:
- 5 Examples
- 6 Please place comments and questions below.
Some definitions that will be used in this proposal.
Please note that these definitions are only being set out in this way for clarity of this proposal and do not necessarily reflect the common usage of these terms in Musicbrainz or in the wider community.
- Release : an overall name for one particular set of Albums and can have one or more entries. For example, a CD single released in several different countries with different track listings would be treated as one Release, as would an original album and a remastered version of the same album (with or without extra tracks).
- Album : a group of single tracks that are combined in a single CD/LP/[other media] format. This includes CD5 or CD3 singles, EPs, mini-discs, 12", 10" or 7" LP records, the Internet (ie: web release), etc. For clarity, disc one of a two disc set would be a single Album under this definition.
- Artist : the person or group that is primarily associated with the Album or Release.
- Data-Set : all the attributes that relate to a single Album, such as the track listing, title, DiscIDs, release date(s), media format (ie CD5, 12", DAT tape), record label, countries released or not released in, the album category, album attributes (ie: bootleg, compilation, re-master, etc), whether it's part of a (box, multi-disc) set and what position it belongs to in that set (if applicable).
ReleaseGroups is essentially a layer between Artist and Album.
The fundamental idea is that a single Release can have multiple Data-Sets attached to it.
For example, the single "Closer" by Nine Inch Nails has several different versions available to it:
- The "Closer to God" single/EP which was released in the US as a single disc with 9 tracks.
- The "Closer: Further Away" and "Closer to God" which was a 2 part single released in the UK, which have 5 and 4 tracks respectively. They were released seperately but were considered part of a 2-disc collecters set.
- The "Closer" radio release promo cd.
- (and I'm sure there are other versions out there, DJ 12" records and so forth)
The idea is to bracket these under the one Release called "Closer". Going into the Release itself would give you more details on the specific Albums available. Each Album would have it's own seperate Data-Set. For each different version of an Album, there would be a new Data-Set.
One fundamantal requirement for this to work correctly is that CDIDs are no longer treated as unique entities and can be associated with multiple "Albums". This will allow a Box Set to be a seperate Release to the single and share the same CDID.
Details of a Data-Set
Specific details of what would be included in a Data-Set. Very few, if any, of these details should be mandatory.
- AlbumTitle - a particular Album can conceivably have a different title to the overall release.
- ReleaseDate - Year, Month, Day of release (as specific as is available).
- RecordLabel - the label that made this particular Album.
- MediaFormat - my basic list so far for this is - CD5, CD3, LP, 12", 10", 7", mini-disc, cassette, digital (meaning a data file, ie: mp3, wave, used for web releases), digital audio tape, 8mm tape, 78rpm record, audio cylinder, film/video...
- TrackListing - this will be expanded to handle multi-side releases side as records.
- ReleaseCountries (released/not released in) - The present list would be expanded to include some non-existent countries (ie: East/West Germany, etc) and also to cover common combinations such as Benelux (Belgium, Netherlands, Luxemberg), Europe, Asia-Pacific, etc. This will also allow for a "not released in" variant, to cater for albums which have "Produced by <record label> for countries outside of the Unites States" or similar on the packaging.
- AlbumCategory - Album, Single, EP, Internet, Other. Please note that this usage of the word "album" is different to the one used for the rest of this proposal and is meant to mean the more general definition of Album - being a group of songs by an Artist or various Artists in a compilation, but not released as a Single... this is one of the confusing double usages in the music industry.
- I'm inclined to drop 'EP' altogether as it's fairly ambigious - as far as I'm concerned, and EP is still an album, it's just a mini-album. Single is fairly well defined. I'm unsure about "Internet" as a Category, this can possibly be covered with "MediaFormat" instead. Other is there to account for things like Software and other releases that might not be an Album.
- AlbumAttributes - potentials are:
- - this Album contains live tracks.
- - this Album contains remixed material.
- - this Album is a compilation or anthology.
- - this Album is a bonus addition to a Release.
- limited edition
- - this Album has been produced in a limited run.
- production run
- - how many copies were produced (mainly for use with limited edition runs)
- - the audio on this Album has been remastered, usually from analogue equipment.
- - this is a later release of an older work. Music that was previously only available on vinyl and then released on CD later would gain this classification; also later releases that have extra or different tracks could also be classed as a re-issue. This is sometimes referred to as a re-release.
- spoken word
- - this Album contains tracks which are spoken word. Audio books would also fall under this classification
- picture disc
- - usually limited to vinyl releases, this refers to an album that has a picture printed over the grooves of the record. They are generally limited edition collector's items.
- special shape
- - this means that the media is cut into a special shape. Like picture disc above, they are generally limited edition collector's items. For example: http://www.atu2.com/collectors/columns/mcgee/ufshaped/. This would be classed as "special shape", "picture disc" and "limited edition".
- - released specifically for promotional purposes by a record label. Generally these will only be made available to DJs and radio stations, but this classification will also cover CDs released on magazines or as giveaways.
- - a release that is not sanctioned by or is released indendently of the artist and/or their record label.
- - this Album is associated with a movie, television show or theatrical performance.
- - this Album is an officially sanctioned release of the artist and/or their record label.
- - the audio was provided as part of a computer game or other software package.
These attributes would be presented as tickboxes and multiple attributes could be applied to one Album.
- Set - this would define whether this Album was part of a set, ie: belongs to a box-set, or is one of a multi-disc set. This would also apply to disc sets where the discs are sold seperately, as quite a number of UK singles are.
- Place in Set - this would define the Album's position in the Set, if applicable. For example, disc 2 of a multi-disc Album would be position 2 in the Set.
Problems to resolve:
There will be some issues regarding cross-talk between a box-set Release which combines other Releases into one,
Migration of all the current meta-data will need to be laid out in specifics once the basic Release scheme is finalised.
There is potentially a great deal of redundancy inherant in this system. One possible solution would be to create a "master" Data-Set, and have all other Data-Sets use a "diff" approach to storage (ie: the Album Title differs from the master set in such-and-such a way, the country data differs in this way, everything else is the same).
Handling of Sets still needs some fleshing out.
Please add examples here, especially a good selection of both typical cases and edge cases.
Multiple variants of an album released in different places or contexts
Closer, by Nine Inch Nails:
- Closer to God, released in the US as a single disc with 9 tracks.
- Closer: Further Away and Closer to God, a 2 part single released in the UK, which have 5 and 4 tracks respectively. They were released seperately but were considered part of a 2-disc collecters set.
- Closer, radio release promo cd.
- Probably various vinyl variations and other stuff
Double CDs and box sets
Best of the Beast, by Iron Maiden - individual discs have numbers, so they should be ordered sensibly:
The Fragile, by Nine Inch Nails - individual discs have names instead of numbers:
Use Your Illusion, by Guns 'n' Roses - double album, but individual discs sold separately:
Collected Recordings, by TISM - a box set of previous TISM releases:
Intelligence and Sacrifice, by Alec Empire - released both as a single disc, and as a two disc set, with the second disc possibly being a bonus disc depending on how you look at it:
Bonus discs and other bonus material
Baptism, by Def FX - released initially with the bonus live disc, later without:
- Live from a Small Planet
St. Anger, by Metallica - includes a bonus DVD:
Please place comments and questions below.
Question: Should how should we handle individual singles with disc numbers? (eg most radiohead singles are released in 2 seperate versions, with a disc number on the cover). We currently don't store this data, but it should be handled by ReleaseGroups. Either the disc number should be added to the ReleaseAlbumTitle, or the discs should be grouped as a ReleaseSet, each with a ReleasePlaceInSet. ZeroGravitas
- I've updated the notes above to spell out that disc sets such as these should be handled as part of the Set/Place in Set attributes. It makes sense to treat them as a Set because the record label specifically released them to be associated with each other. TarragonAllen
I fully agree with this approach in regards to box sets and other multi-album releases (official or bootleg). However, I think this would be a bad approach for dealing with the multiple releases over time of a specific album. Re-releases of an album can be distinguished by release date/country, and often have differing track listings (i.e. different number of tracks, or different order to tracks). But having a single entry for an album name representing all the different releases is, well, non-intuitive. KISS.
Similarly, I feel the media format attribute is relatively pointless. There's a huge number of formats, and they will continue to grow. But the format has little bearing on the reset of the content. Again, KISS.
Regarding the handling of the same album being released with different names, I think there's better solutions possible. Extending the ReleaseCountry attribute to include an alternative title, for example. Or for that matter, AlbumAnnotation.
For there to be any hope to achieve this proposal, it needs to be as simple as possible to minimize the number of changes needed to the DB and GUI, as well as minimizing the amount of code to develop. If planned logically enough it can, however, be extended as needed.
--Jeff The Riffer (aka riffer)
- I think of this in terms of navigating through the website. For many prolific and popular artists, their album list is horribly confusing, with each album appearing multiple times, all jumbled up together. It's often quite hard to answer the question "how many albums has artist X recorded?", because the same album gets released, re-released, remixed, renamed, and resold over and over again, and the different variants all enter the database separately. What I want to see is a simple list of the actual ten or so albums that they recorded, with the ability to then click on each album and get a more detailed listing of the regional variations, the singles that got released from the albums, and all the other stuff. I also think that recording the format is incredibly important. It helps to distinguish *why* there are three versions of every album (one CD, one vinyl, one cassette) with identical track listings and identical release information. It's also extremely important if I'm going to buy the album! If I click on the cover art and go through Amazon's ordering system on auto-pilot, I don't want to get a cassette in the mail. --MatthewExon
It seems to me that there are (at least) two completely different problems here - the problem of representing box sets, and the problem of representing different versions of what is basically the same product. The distinction is, what gets sold independently? Use Your Illusion is two separate products, while Intelligence and Sacrifice is one. There are several different products that are variations on Intelligence and Sacrifice, some with two discs, some with only one. BarCodes and release dates are attached to products, not discs. Since we don't have any concept of a "product" in MB at the moment, there's nowhere appropriate to attach this data. Meanwhile reviews, annotations, and performance credits belong at the "recording" or release group level, which is also a concept we don't have in MB at the moment.
Both are important. I think they are sufficiently different that they require independent technical solutions. They should probably be handled as separate projects, one after the other. This would also help limit the scope, so that we can avoid getting bogged down in weighing up too many independent requirements.
The AlbumCategory and AlbumAttributes above don't always look like they belong at the same level. Attributes like "promotional" or "limited edition" sound like they belong to the product as a whole, not to an individual "data set" (disc) included in it. Others, like "soundtrack" or "live" could apply either to the product as a whole, or only to individual discs in a multi-disc package.