Duplicate Artists

[jgdye]

I'm getting a handful of duplicate artists, where one or two albums are listed under a generic id, and the rest are listed under the MBID. All the files have the correct MBID. I've attached eyeD3 outputs from two files. This one shows up under the proper MBID. This one shows up under a generic id. The difference that I've been able to tell is that the ones with generic ids have both the all caps MUSICBRAINZ_* and "TXX:MusicBrainz *" tags. The problem seems to go away if I delete the MUSICBRAINZ style tags.

[epoupon]

Hello! Thanks for reporting What lms version do you use?

[epoupon]

And could you please paste the full result of lms-metadata on each file?

[jgdye]

Working MBID Non-working MBID I was on v3.43.0, but I just compiled v3.49.0 with unchanged results, and that's the version the lms-metadata outputs are from.

[epoupon]

According to what you sent, it looks like TagLib concatenates the values found in both "MUSICBRAINZ_ARTISTID" and "MusicBrainz Artist Id" tags: Here is what Taglib reports (note the SEP added by lms-metadata to show multiple values): [MUSICBRAINZ_ARTISTID] = e69db8ef-ffd1-4916-a85f-df428f223cce*SEP*e69db8ef-ffd1-4916-a85f-df428f223cce LMS tries to perfectly match the artist name and the artist MBID. Here we have a single name for two mbids => no MBID match is even attempted. For this particular case we could work around this by removing duplicated values.

[epoupon]

Would you mind sending your offending file @ [email protected] ?

[jgdye]

That's what it looked like to me when I scanned over the output.  I would suggest comparing them, using the value if they are the same, and using a generated ID and logging it if different.  Picard at least uses the "MusicBrainz Artist Id" tag when tagging, so just using one of the values could cause a file that had the artist changed at some point to show the wrong artist even though Picard and other taggers show the right one.

I sent the file as requested.

February 8, 2024 at 3:02 AM, "Emeric POUPON" @.***> wrote:

According to what you sent, it looks like TagLib concatenates the values found in both "MUSICBRAINZ_ARTISTID" and "MusicBrainz Artist Id" tags: Here is what Taglib reports (note the SEP added by lms-metadata to show multiple values): [MUSICBRAINZ_ARTISTID] = e69db8ef-ffd1-4916-a85f-df428f223cceSEPe69db8ef-ffd1-4916-a85f-df428f223cce LMS tries to perfectly match the artist name and the artist MBID. Here we have a single name for two mbids => no MBID match is even attempted. For this particular case we could work around this by removing duplicated values.

— Reply to this email directly, view it on GitHub https://github.com/epoupon/lms/issues/415#issuecomment-1933541571 , or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDGTDEJQOHGEZ5WDWIQSMLYSSBALAVCNFSM6AAAAABC6K2S5WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZTGU2DCNJXGE . You are receiving this because you authored the thread.

[epoupon]

Ok, I think I may have a workaround for this issue. But I am not that confident. Will have to test more