Why don’t Classical releases have track artists?

Tags: #<Tag:0x00007f34243c8750>


Classical (in the broad sense) has composers, arrangers, librettists, and various performers; it does not have recording artists because, well, it predates music recording by a few centuries (well, except of course sheet music is a way to record music, but yeah…)

Calling someone(s) the “artist” is just to make classical fit in with things designed for pop; if you were building a database just for classical, you’d probably not have “artist” fields at all. At least except for some very recent composers, depending on how broadly we take classical.

As a side note, a lot of classical releases don’t even really have albums. Look at the title—it’s just a list of works on the album—and those names are often just numbering them (Symphony no. 1, etc.) and sometimes not even assigned by the composer, but rather by musicologists digging through his papers after his death. The selection of what goes on the disc generally isn’t done by the musicians responsible for writing or performing the music.



To show some example artwork:

It has a set of artists for the entire release, but no track-specific artists at all. This is pretty much the same as a normal pop album, except that putting the entire artist set on each track tends to get unwieldy.

Similarly, http://www.hyperion-records.co.uk/notes/68103-B.pdf (tracklist on page 4 of the PDF) – global composer and performer; performer group artist is broken out into four singers; singers identified on each track by initials. I would not count those sets of initials as “track artists” – the way the release seems to de-emphasize the individual singers on any given track suggests this is also release artists only (composer + performer group)

http://www.hyperion-records.co.uk/notes/67929-B.pdf (page 3) – global performer; there are not per-track artists at all, but “track group” artists which are exclusively the composer (apart from one track group which lists a single guest soloist). Since MB can’t give credits for a group of tracks, the best we can do to match the release is list the composer on each track.

http://www.hyperion-records.co.uk/notes/67997-B.pdf (page 3) – global performer, two release composers; tracks broken into two groups by composer. Just as above.

http://www.hyperion-records.co.uk/notes/68087-B.pdf (page 3) - global performer. This is the first one that seems to arguably have track credits beyond the composer. It does list the composer or a similar source as well as arranger and soloist, but the format (and de-emphasis compared to the global performer) suggests to me that these are more akin to our ARs rather than track artists.
Consider track 6: “Owain Park / mezzo soprano Helen Charlston, soprano solo Faith Waddell” – CSG style would say “Owain Park”. Would you have it “Owain Park, Helen Charlston, Faith Waddell”? That lists all the artists that the release lists on the corresponding track but it exclude important performers: the choir and director Stephen Layton. So maybe “Owain Park, Helen Charlston, Faith Waddell, Choir of Trinity College Cambridge, Stephen Layton”. I guess that’s complete but it sure is long and ugly. Now we have a huge list of artists with no indication of their roles so we have to look at the ARs to find them. Since we have to do that anyway, why stuff them all into the artist field when we could just look at the AR list anyway.

This is pretty typical stuff for classical in my experience, but just to show that I’m not only considering one label:

http://www.naxos.com/SharedFiles/pdf/rear/8.573496r.pdf – global performer, tracks grouped by composer, which is the only sort of “per-track” artist.

http://www.naxos.com/SharedFiles/pdf/rear/8.573566r.pdf – global performer, per-track artists are exclusively the composer.


See also http://tickets.musicbrainz.org/browse/MBS-7490


I just would like to remind that all our tracks do have track artist credits. It is just a data display feature that when all tracks have the same AC as the release on a medium, the track AC column is hidden.


ARs dont let you specify a different name that is what Im saying. I kinow yo can specify this detail as recoridng or track artist credit that is my point.


Maybe this is the key thing, the word ‘artist’. So yes okay the word ‘artist’ is not the best word for Classical, however ‘people’ are credited at release and track level, if within MusicBrainz you consider ‘artist’ to mean person then thats not a problem, in fact we already do this as every person in the MusicBrainz db is in the Artist table. Composers thought in terms of works but we still buy and consume music as ‘albums’, that is why in this thread How to explain Works I want to add works to tagging metadata to allow users to browse by work rather than album


They do; you may not be up to date on the MB schema if you are convinced otherwise.


Thanks for taking the time to do these examples

Yes this is an example of bog standard classical release, as you say like many pop albums there is no difference between the release and track credits. So my proposal here would the recording artist credit would be essentially the same as the release artist credit with composer, orchestra, conductor. Track Artist credit would be exactly the same as recording artist credit (at database level would point to the same artist_Credit record.

Agreed, so I would have composer and performer group for release artists/track artist/recordng artist. Then the individual performers would be added via AR’s, this allow differentition between who actually performeed (ARs) and who is given credited on the cover (composer + performer group)

Yes you could call them “track group” or maybe “work artists” as typically differences is at work level. But you say the best thing you can do is put the composer at track level, I would say the best thing you can do is put the composer at track level except tracks 6-11 where you also put the percussionist. To me the way the composer and soloist are credited at group level rather than track level is simply to make it most easy to understand for the consumer, there is no point repeating the composer for every track in a work as clearly (though perhaps there are some edge cases) all tracks from one work will be composer by the same composer, and most likely the solist will also be present an engaged throughout the work.

As I assume Stephhn Layout and choir are still involved as release artists would say "Choir of Trinity College Cambridge, Stephen Layton, Owain Park, Helen Charlston, Faith Waddell"Just like Pop/Rock I would like to differentiate between who is credited and everyone involved. So the as track artist/recording artist that tells us who is credited then we can use performers AR to allow addition of all the people, including any extra people that were involved but not deemed sufficiently important to be given main credits. If a RockPop track was credited to six people you wouldn’t expect to not list them all because it looks ugly

So I think now I understand what you mean by there being no track artists, its essentially that there is unlikely to be a single track that is credited differently, tracks are more likely to be credited at work level (what about one track works), which may or may not be the same as release level, but as your own example shows sometimes credits are at track level .

To me the fundamental differences between Pop/Rock and Classical are really that:

  1. Multiple people are credited for each release/track, always composer, usually conductor and orchestra and sometimes individual musicians.
  2. Credits natural grouping is at work level rather than album level

I’m sure there are examples of Rock/Pop releases that also credit songs at a ‘group level’ rather than a track level. In theory you could extend MusicBrainz to group level credits but I dont really see it would be of much interest. What I don’t see are good reasons for why track artist credits and recording artist credits mean different things in CSG, it seems better to me to simply treat group level credits as track artists credits for the tracks involved which Im sure is what we would do for Rock/Pop with group level credits.


I was basing it on the UI, I didnt look at the schema but as far as I can see you cannot do it when you add a relationship on the site.


There is a text field labelled As, in the relationship editor pop up, that is made for artist credit on the relationship.


Ah, didnt notice that thankyou.


Historical reasons mostly.

  1. Pre-NGS we could only have one artist per track.
  2. Most important artist was deemed to be the composer.
  3. Thus CSG track artists were set to composer.
  4. Recording artists were copied from tracks, so recording artists were also composer.
  5. This made classical recording artists useless (due to thousands of recordings now appearing to be “by Bach” for example) and illogical (in many/most cases the composer had no direct involvement in the creation of the audio recording)
  6. There were previously no standards for recording artists, so this was defined to be performer to fix the above.

Maybe something is different now and another change is warranted. I would support relaxing the requirement that track artist only/always be the composer, but I don’t think I would go so far as to list every artist that happens to be mentioned; this is for convenience of data entry as well as cleanliness of display.

I think that for many classical releases (including the previously-mentioned “Yulefest!” release), the very detailed “track artists” are basically what we would normally set with ARs. They’re not proper track artists. This also applies to that single track with the guest soloist. Compare to the pop music situation – have a look at https://ia800308.us.archive.org/9/items/mbid-cf0d899c-bbc6-4a33-ba74-5e335284e836/mbid-cf0d899c-bbc6-4a33-ba74-5e335284e836-1353923836.jpg
– we wouldn’t consider the “(Barrett)” text to be a track artist, even though it’s directly on the tracklisting.

I would not mind if, in cases where there was no specific track artist listed, we would consider it “obviously they meant that the release artist should apply here” and entered it matching the release artist, as we do for pop. This has the advantage that if the release is all the same artist the display will not bother repeating that information for every track. However it falls down as soon as a single track is different in any way.


I’m sure I’m not alone here, but I’ve been losing the thread a little bit! (again)

It would be really nice to see a clear and concise proposal, perhaps on @ijabz wiki page (eg not somewhere where it can be buried) as to what different guidelines would be proposed, and also some examples as to how they would deal with the examples given by Hawke (showing how the new credits would look vs the old). It’s probably already been written down somewhere, but I can’t remember!

I understand the ideological argument happening but my brain is starting to fray a bit visualizing the details :sweat_smile:
Personally a ‘classical’ tag or checkbox would be a pretty huge step forward, especially if/once Picard supports it.


Nothing to disagree with you about here, I think you are probably right that those musicians on YuleFest would just be set with ARs, For a soloist to be included as part of the track artist credit possibly would need to be on the front cover.
So to summarize it as I see it:

  1. Treat Recording Artist Credit and Track Artist Credit as being the same thing (like we do for non classical), so the Track Artist credit is only different to the Recording Artist Credit if the particular recording is credited differently on a release to how it was on the first release it was part of.

  2. Track/Recording Artist Credit would represent the credits on the cover (like we do for non classical) in the majority of cases be the same as Release Artist Credit.

  3. The most likely exceptions would be cases such as when one release has two concertos by two different composers, in this case the release artist credit would list include both composers but the track credits for tracks belong to concerto 1 would list just that composer, and for concerto 2 just that composer.

  4. Like Release Artist Credits Track/Recording Artist Credits would typically for classical contain Composer(s), Soloist(s), Conductor(s), and Orchestra(s) but of course it depends on the release.

  5. Adding a new release should be easier. With the current guidelines you usually have to edit the track artist credit so only contains composer, and then have to manually edit each recording artist credit to contain only performers. With the proposal, because when adding a new release if there is no difference between track artist credits and release artist credit the track/recording artist credit does not need editing. If there is a difference then the track artist credit will need editing BUT even then what you wont have to do is separately edit the recording artist credits afterwards.

  6. Maintaining the database would be easier, there are many releases where the CSG has been followed for Track Artist Credit but not Recording Artist Credit, this would not be an issue with this proposal.

  7. Interpreting the database would be easier. Currently you need to try and guess if the CSG is being used in order to interpret track/recording artist credits differently. With the proposal you’ll get decent results without doing this, although I still think the CSG flag has some merit because of other differences between CSG releases and non CSG releases.

To summarize my main issue with the CSG was data integrity issues this would resolve it. But Im no Classical Music Expert so creating comprehensive examples might be better done by someone else, if there is anyone else with interest in this idea.


Curious what you suggest in the cases where the only credit on the cover is the composer, which is reasonably common, or where there are no credits on the cover. Presumably you’d then use the composer, but that’s a problem on the recording—or you make an exception there, which would seem to bring back your data integrity issues.

(Also, it’s quite common to put works recorded by completely different people—different orchestra, different conductor—on the same release.)

PS: The length of the album credits already make problems frequently enough (by exceeding file name length limits). Having that same thing as a track artist credit should be interesting, since in a lot of common tagging schemes that gets combined with the track name, which also can be fairly long on classical releases.


I object to this portion. IMO recording artist needs to be the performers, regardless of what the track listing might say.

Consider http://musicbrainz.org/work/0d2a5e67-8efc-3233-9ac9-f4be8644be14 and note how many times the recording artist is simply the composer, who had nothing to do with the creation of the recording.

I think even with your proposed changes we will find a fairly large number of instances where the tracklisting does indeed just include the composer, and if recording artist matches that it will be very hard to distinguish recordings.


But they can be added as advanced relationships, I don’t see the need to add them as ARs and then have to add them to recording artist as well if don’t have a track credit apart from buried within the release booklet. Performers who don’t merit a track credit can still be searched and all their relationships are available in their relationship tab.

I would think in alot of cases the recording artist is simply set to the composer because whoever added the release followed the CSG for track artist but didn’t bother doing the more awkward recording credit. Or the didn’t use CSG and if only the composer is credited that is what was added, but if the performer was not credited then why should they have a recording credit.


There is no real problem because both releases contain recording on many releases. If the recording only appears once on one release and the performers are not credited so be it, but if they are credited on multiple releases and on one of these releases they are properly credited (which is to be expected on a non-compilation) then they can be properly credited because in the normal Pop/Rock case recording artist credit contains the first/best credit for the recording and the track artists credits are the same as the Recording Artist Credit unless they need to reflect something different on that particular release.

This isn’t really an issue for MusicBrainz, its not a good way to design a db based on filesystem limits on end users. But taggers should cope with it, and its less problematic for artist credits then titles anyway because artist credits are represented in the webservice as separate artists not a single string like the title.


Because the composer had nothing to do with the creation of the recording, in the vast majority of cases for classical music?

The point is that while either way you end up with recordings that have only composers in the artist field, at least with current CSG it’s wrong and can be fixed so the system is useful.


So basically: “treat the track artist field exactly as printed on cover/ exactly like with pop releases”

I’m not really convinced, extreme variation between gigantic and extremely short or irrelevant (or wrong) artist credits seems to be fairly normal for classical, and both situations throw up a pretty large number of annoying problems (both in terms of browsing the site and tagging files) that I think are easily and consistently solved using ARs.

I think the flag is important, and another avenue to explore might be to suggest some kind of variation on the ‘alternate tracklists’ suggestion (http://tickets.musicbrainz.org/browse/MBS-4501) that allows both a pop-style and a classical-style artist credit to be entered? Would be slow going, but it’s a thought.