Picard v3 `albumartists` & diff highlighting question

I have recently upgraded to Picard v3. Looks quite good overall. But I just loaded an already tagged album and I have two questions, which I found no answer to anywhere else.

First, it seems Picard now inserts both albumartist and albumartists tags. What is the purpose and thinking behind the new albumartists and how it will be different from a singular albumartist? Both seem to contains the same information.

Second, all 3 tracks on this album are missing the new albumartists tag, however the diff highlighter (red) is brighter on 2 of them and completely missing on the last track.

Why is it the case? I was relying on the differ a lot to see potentially problematic tag overrides and if it starts to misbehave it loses trust and forces me to check every single file manually.

Thank you for your help.

There is only a difference when there are multiple artists. “Album Artist” has the artist names as credited in a single string, with joiners and all, e.g. 'Neil Young with Crazy Horse'. “Album Artists” has the individual artists in a multi-value tag, i.e. ['Neil Young'; 'Crazy Horse']. It’s the same as with “Artist” and “Artists”, which you both already have.

Regarding the diff indicator, you would have to inspect the details in the metadata pane. Maybe those recordings have been merged with others and got a new MBID, or a composer credit has been added. Or it might simply be that “Sunflowers” has 10 letters while “Smile” has only 4, and the longer match in the title outweighs the missing Album Artists tag.

1 Like

So this ‘Album Artist’ addition is specific to Picard v3? It is suprising that it wasn’t mentioned anywhere, as this is a relatively major change (now an entire music collection is in a dirty state as it wants to insert this new tag).


I have inspected the metadata. This is the only field that’s changing. The title length also doesn’t hold true, here’s another example from a different album where some tracks even with a very long names are marked red, while others aren’t.

It is not prominently announced as a new feature, but it is noted in the changelog. If that new tag is of no use for you, you can simply add a one line script and $unset(albumartists).

Regarding the diff indicator I was only guessing. I noticed that for some of my albums, too, but I didn’t look into it any further.

1 Like

Ah, so it turns out the Changelog for the public 3.0 release didn’t include entries from release candidates, this is how I missed it :person_facepalming: I think it should have been included.

Thank you for pointing out.

I will report the differ issue in Jira.

It’s written by default (not behind an option), matching how “artists” already behaves. The main reason is that the players people actually use for this not only read the tag but recommend writing it:

  • Kodi reads ALBUMARTISTS (Vorbis/ID3/APE/MP4) and its wiki explicitly tells users to add a script to write it, because Picard doesn’t do it yet.
  • Navidrome maps albumartists directly and prefers it over splitting the single albumartist string.
  • Jellyfin reads it too, when the “Prefer nonstandard artists tag” library option is enabled (otherwise it splits the singular tag).
2 Likes

I chuckled a bit when I saw the ticket number in the changelog. “Wow, 700 - that must be old” :grinning_face_with_smiling_eyes:

It’s a pity that there are still no multivalue sort tags. But I found the Additional Artists Variables plugin which has all the information I could ever want, so I am not complaining.

It’s available as a variable:
$setmulti(albumartistssort,%_albumartists_sort%)

1 Like

Oh, indeed. I totally overlooked that, thanks!

In addition to the change log as mentioned earlier, it is identified in the documentation as being new with Picard version 3.0.

2 Likes

Thanks everyone for the context about album artists tag.

Regarding differ, I went ahead and created a bug report: https://tickets.metabrainz.org/browse/PICARD-3527

1 Like

I tried to address the inconsistency in PICARD-3527: Make the track match differ reflect actual changes by zas · Pull Request #3535 · metabrainz/picard · GitHub

I think it is much more intuitive now, and somehow similar to Picard 2 (actually not totally because Picard 2 behavior was actually due to old bugs we fixed in Picard 3).

The icon was expressing “the quality of match between the information from the file and the information for the track as provided from the MusicBrainz database” but now it is clearly “how closely the data currently loaded in the file matches the data stored on disk”.
This seems to be the behavior most users expected, and it is much more useful.

I also open a PR for clarifying documentation.

If someone can test from source, and confirm, it would be great.

1 Like

Thank you for such a quick turnaround.

I cannot build from source, but happy to test once this is on beta.