Accessing extended performer metadata from Musicbrainz

[CrimzonCoder]

For classical music, MusicBrainz policy specifies that the Artist tag have the composer and all of the performers are stored in PERFORMER extended metadata. The autotagger doesn't pull that information, but even if I add the metadata to the files using mp3tag, I cannot figure out how to extract that information.

PERFORMER: Yo-yo Ma (cello)
PERFORMER: Berliner Philharmoniker (orchestra)

It seems like I should be able to get this using python and the inline plugin, but alas...

Is there a plugin that handles this already? If not, how can I use python code in the inline plugin to get at this data?

[keremoktener]

For tags you've already added to the files, inline can read them with Mutagen, which beets already depends on. Inline's normal variables come from the beets item, so an arbitrary file tag such as PERFORMER won't automatically appear there.

Add inline to your existing plugins list, then add this top-level config block:

item_fields:
  performers: |
    import mutagen
    from beets.util import syspath
    audio = mutagen.File(syspath(path), easy=True)
    if audio is None or audio.tags is None:
        return ""
    values = list(audio.get("performer", []))
    for key in sorted(audio.keys()):
        if key.startswith("performer:"):
            role = key.split(":", 1)[1]
            values.extend(f"{name} ({role})" for name in audio[key])
    return "; ".join(values)

Check it without changing your files:

beet ls -f '$title: $performers'

Checked on beets 2.14.1 with synthetic files covering FLAC PERFORMER, MP3 TXXX:PERFORMER, MP3 TMCL musician credits, and missing tags. TMCL is why the snippet also reads performer:<role> keys.

This creates a computed $performers field for templates/list output. It doesn't persist the field in the database or fetch missing MusicBrainz relationships. If your files are M4A or another format, which format are you using? The tag mapping can differ.

References: inline block syntax, Mutagen musician-credit handling, PERFORMER mapping.

[CrimzonCoder]

This code worked like a charm. And the references were very helpful. Thanks!

[aashish254]

for classical performer roles, the inline plugin is the right tool — it lets you use python expressions in path formats against any field beets knows.

first check what beets actually imported: run beet ls -f '$performer' on one album and see how much role detail came through. musicbrainz sends performer relationships, but beets only maps a subset by default.

if the role detail (soloist, conductor, orchestra) isnt in the stock fields, youve got two options: a tiny custom plugin that reads the musicbrainz recording relations during import, or the fromfilename fallback. the data exists in the mb api response — its just not all mapped to beets fields out of the box.

paste what beet ls -f shows for one track and i can tell you which fields to use in the inline expression.

[CrimzonCoder]

None of the performer information is imported by the beets tagger. Even when I add the extended performer data using another tool, those tags are not mapped to any tags in the beet database.

I will either have to add the performer information manually using another tool as I did for my test cases or, as you suggested, add/modify a plugin to include the desired tags, and perhaps add them as valid fields in beets as well.