The "When is it a soundtrack / when is it classical?" question

Tags: #<Tag:0x00007f050c8d93a8> #<Tag:0x00007f050c8d8e80> #<Tag:0x00007f050c8d8c50> #<Tag:0x00007f050c8d8890>


Haydn’s music played by Schiff with Schiff lecturing.


In that case I’d definitely use CSG.


Normally with video game covers, it’s the artist doing the arrangement/performance that gets the credit


I just ran into a “Performer” vs “Composer” problem:

Jan Garbarek & The Hilliard Ensemble: Officium Novum

and I thought to myself… gee, it would be nice if there was a “classical” flag on Musicbrainz that Picard could read, in combination with a user preference, so I could tell it to tag my classical CDs in the “pop” format (i.e. performer as artist vs composer as artist).

It might also discourage new users from attempting to “fix” classical releases, like I did in the edit above.

Than I searched and found this thread. So yeah, I support this idea wholeheartedly.

In the meantime, does anyone know of a Picard script that I could use on classical releases to “pop-ify” them? (i.e. put the composers in the composer field, and put the performers in the artist field)

[Disney] or Composers for Disney track artists?
[Disney] or Composers for Disney track artists?

I thought about this some more and the biggest problem is simply that the CSG says to use track artist for composer. I dont see a need to do this since we have the COMPOSER relationship.

Not only does putting the composer as track artist makes it very difficult to decode the recording artist/track artist since there is no flag to indicate that the CSG guidelines are being used as already discussed.

But using the same field to represent different things is a very a bad design decision

Furthemore all the valid reasons for having difference between recording and track artist for Popular music cannot be done for classical music because of the requirement to use track artist for the composer. i.e if a performer of a recording is credited slightly differently on different release we cant capture that.

And am I right in thinking Picard use to have an option to use recording artist instead of track artist but has since been removed ?


I very much agree here, I never completely was convinced that special handling is necessary. One argument that will come up though is the user interface issue on If one looks at the artist page of a classical composer, one very much expects to find what he wrote. But since we have works now for quite a long time, and producing works is what a composer does, I don’t see this issue as important anymore.

Not that I can remember. And also it would again be the problem that we are not able to determine programmatically if CSG applies or not. Does Jaikoz have some special handling here?


Yes the works tab for an artist (who is a composer) does seem to do the job now.

Not currently but as a starting point you could:

  1. Simply give option to user to always use recording artist credit instead of track artist credit. For most Pop/Rock songs it would make no difference (though it would for some) but would be a big improvement for Classical (unless you wanted composer as track artist)

  2. When getting metadata compare track artist credit and recording artist credit for track.
    1st check: If the number of track_artist!=recording_artists good chance of being Classical and use Recording Artist
    2nd Check:If the artistids are different (not credit name coz that can be different) then we have a different list so probably Classical and use Recording Artist.
    Possibly only return Recording Artist if 1st or 2nd check true and if there is a composer relationship the composer id matches the track artist id.


I guess one reason we didn’t rely an taggers using COMPOSER before is because a lot of players and media centers still do not really support the COMPOSER tag. Kodi only just got support for it earlier this year. I’m guessing a lot of smart phone music players also do not support it.


lol, the mantra used by MusicBrainz so often is that it is a music meta database with many uses not just tagging. But now you are using the metadata deficiencies of some software to justify this bad decision - I don’t think this was really the reason anyway


No matter which way we go, we will need a way to differentiate between classical / non-classical releases. I can see the reason why people want to have the composer as track artist in their files very clearly, as even for players supporting composers the performing artist is the primary thing displayed. So to satisfy this tagger must have an option to write the data both ways and therefore taggers need some clue when to apply which logic.


The reason for this is that people generally like their tags to match what the cover art says and this is not always the performer – in the case of classical it is rarely the performer. Maybe the CSG should be updated to say essentially “use what’s on the cover, but the cover doesn’t give a good answer, use the composer instead”.

Of course that wouldn’t help at all in the case of this “Officium Novum” since the release does give track-specific artist credits.

I don’t even know why this is an issue: Any tagger could normalize this if they wanted. The tagger is able to determine whether the artist in the ‘track artist’ field is composer, performer, or really anything else. If someone wants their music to always use the performer in the artist field, that’s up to the tagger to do. To my knowledge there has never been much call for this, for classical or otherwise.


First Freso then you, when did we start defining how MusicBrainz should work base do what taggers want ?
Anyway when I look at my classical records they do typically specify Composer, Conductor and Orchestra, Soloists at track level if it varies between tracks and if they don’t they specify composer/conductor and orchestra for the release as a whole they dont just specify the composer. And I’ve spoken to serious classical music enthusiasts and they want to see performers in the track artist field so they can more easily differentiate between different performances of the same work

Exactly so that clearly demonstrates why the current standard makes little sense

Yes we can try to do this but its not straightforward it depends on what data is available, its certainly not possible to do it 100% reliably.

But what we clearly do have is a composer relationship so what is the point of storing the composer at track artist credit and making the field mean different things for different types of release.


Composer relationship can be used only with works. It can be extremely hard to identify correct works for non experienced editors. Because of this we usually see new releases without recordings linked with works. Another problem we already got is growing amount of work duplicates because editors haven’t been able to find correct works.

I believe most of the serious classical music enthusiasts would differentiate performances by using relationships with dates. It’s quite common for conductors to record same works multiple times. Often with the same orchestra.

Track artist field is vague field without a definition but I don’t have a problem with that. It isn’t any way different than what we already see on releases. There’s surprisingly many (low-budget) releases not crediting any performers or not crediting the roles of artists. We can always use relationships when there’s a need to define the roles.

I’m quite happy how things currently are because current style is also easier for editors adding new releases. Most of the releases are added by editors who don’t care or know about relationships but just add releases with minimum amount of data. With compilations it would mean from 2 to 3 times more data to add (if using performers as track artists) and would easily require adding tens of new artists.


Surely a work for each recording should be a prequisite for classical releases

Well no because they should be adding them as the recording artist credit anyway. If they are not then the information that is required is not being added to the release. And the relationships with dates wont be there either


So which way do you want it? Do you want the data in Musicbrainz to reflect what taggers need (since they can’t apparently do it from the relations alone?) Or do you want taggers to be required to use the relations and perhaps do a poor job from it?

But it doesn’t. That’s what I said above: The artist field means “whatever the release says the artist is”. In the case of Officium Novum that will continue to be the composer no matter how you change the guidelines. I mean I guess it would be possible to change the guidelines to “the track artist will always be the same as the release artist” but that seems a huge step backwards.

Why is it any better to consistently have “track artist = performer” than the current method?


For non CSG Track Artist Credit is a refinement of Recording Artist, but for CSG Track Artist has nothing to do with Recording Artist Credit - it stores completely different information. That it uses the field in a completely different way so preventing it usual use (i.e an Orchestra credited differently for the same recording on two releases) and doesn’t even give any indication that it is doing this ((via a CSG checkbox) is the key issue.

I don’t understand how you cant see that using the same fields in very different and inconsistent ways is not a problem.
Its analogous to using the Barcode field to store the Vinyl colour for coloured vinyl for releases that don’t have a barcode and just mentioning it in the style guidelines


We would have a huge problem if only place to credit composers would be on work level. Recordings without linked works would be impossible to fix later. How can you identify “Violin Concerto: I. Allegro” if there’s no info about the composer? It’s already pretty common that we aren’t able to identify correct works because title doesn’t include any catalog numbers. Where should we put information about the composer if we don’t know the correct work?

Reality is that people don’t add all the available data and there’s no way to force that. If we have a guideline which recommends adding performers as track artists they will most likely do only that. Current situation isn’t ideal but we can survive. We often know the performers if release artists are credited and still know the composers. Reality could be that we would end up knowing only performers but have no idea about the composers.


The composer is also usually part of the release album artist and I would not be adverse to adding the composer to the recording artist credit. But as I keep saying, (and keep getting ignored in replies) the issue for me is the way we use the trackartist field in a completely different way for some releases and we dont even indicate which releases we do this for.


and why when you look at a recording does it have a column for the release artist but not the track artist !


As I think I’ve said, I do like the idea of being able to flag a release as being ‘CSG’ or ‘Theatre Style’ or whatever.

But is the track artist field being used inconsistently? It depends on what you think the correct usage is. One interpretation (as @Hawke said) is: the track artist field says nothing about an artist’s role on the track. It merely indicates that this is the artist credited for the track.

What does it mean to be “credited for a track”? Unfortunately, this seems to be a matter of interpretation. Track lists in the wild are not written to be machine-parsed.

addendum: looking at the default Picard mappings, the artist field maps like this:

  • id3v2: TPE1 - Lead performer(s)/Soloist(s)
  • Vorbis: Artist - The artist generally considered responsible for the work. In popular music this is usually the performing band or singer. For classical music it would be the composer. For an audio book it would be the author of the original text.
  • apev2: Artist - Performing artist | List of Performing Artists