duplicate music artists or loss of overview data

[quixotic120]

scanning my music back in slowly but surely, after rebuilding my entire library for 10.11. all music is tagged per musicbrainz with beets.io. I have only scanned in A and symbols so far.

I have encountered an issue with music library. If I scan with default library settings (musicbrainz only for scraping metadata and audiodb) I am not able to scrape overview data, resulting in entries like this:



solely having the genres and a link to the musicbrainz ID. If I do this the library scans in correctly with no duplicate artists though.

if add audiodb for metadata scraping I get overview and a more robust display:



However, if I do this certain artists appear as duplicates:



this has two entries - the first is the former musicbrainz only (first screenshot) and the second is the full one (second screenshot).

of note:
the first entry (musicbrainz only) points to /config/data/metadata/artists/A-Trak with musicbrainz ID 47ea59b7-81c2-4001-8ac3-567877b3172a
the second (both) points to /data/music/A‐Trak (correct path) with musicbrainz ID 47ea59b7-81c2-4001-8ac3-567877b3172a (same ID) and audiodb artist ID 116298

also of note: only artists impacted are ones with a hyphen in the name (per third screenshot - A-Trak, At the Drive-In) and J-pop where entry scraping both with the correct path has the romanized name (Atarashii Gakko!!) and the other with musicbrainz only and the incorrect path having the Japanese name (新しい学校のリーダーズ). The last artist (Yukarin Channel) in that screenshot is similar situation to the j-pop but it does not have an audiodb entry so it only has one entry. 

deleting the directory in /config/data/metadata/artists/ does not fix this, it is recreated on a library scan but loses images (thus the entry for At the Drive-In that does not have an artist image).

this metadata confuses some clients - these artists don't display correctly in kodi using jellyfin for kodi, for example

for reference this is viewing under the album artist sort. viewing just Artist goes to the correct entry and no duplicates exist, but this view is much messier

So essentially is there a happy medium here? Can I scrape just musicbrainz and get the overview, as most artists have a section of overview (wikipedia data), not sure if this is possible. Or if I keep the audiodb scraper to get things like overview data can I change something to make this work? using the audiodb only won't work because it's not as comprehensive of a library (especially for non-english artists)

edit: two more observations: I see that in the metadata manager in the admin console only the "correct" entries (that reference the correct path) exist.

additionally upon adding the "b" artists into my library I saw one that was tagged incorrectly resulting in 2 entries. fixed this, rescanned library, and the second entry disappeared (based on a featured artist in album artist field). However, going into the offending artists from the screenshots above and ensuring all are tagged correctly with the same artist name for all albums does not fix anything. likely because the secondary entries for the aforementioned albums introduced a semicolon and additional artist (e.g. artist x ft artist y created a secondary entry of "artist x ft artist y; artist x" and changing the tag to remove the ft artist rescraped correctly)

[quixotic120]

scanning in more artists and I have found the following pattern:

If an artists exists on theaudiodb and has a non alphanumeric character in the name (e.g. At the Drive-In, Christopher O'Reilly) there is a duplicate entry made. If the artist exists on theaudiodb and has a primary non English name (e.g. a j-pop artist with a hiragana name and a romanized name) a duplicate entry will be made.

If the artist has those offending features but only exists on musicbrainz, there will only be one entry.

still have not found a fix. If I delete the library and metadata entries (under /metadata/Artists and /metadata/people) they are simply recreated upon recreating the library. Any ideas to fix would be appreciated.

[Dex Luther]

I don't know if this is related, but I'm getting the double artists too. One seems fine, the other is blank.

The one on the left is wrong.


I wanted to know what path they were using, so I could track down the issue, and found this:

The blank one is pulling from the metadata folder inside Jellyfin's data folder for some reason



The other artist comes from the correct path, where the music files are like all the other entries:

[quixotic120]

That is the same presentation as my issue but I have not found a fix other than only using musicbrainz to scrape

[Dex Luther]

That is the same presentation as my issue but I have not found a fix other than only using musicbrainz to scrape

I'm getting it whether scraping from Musicbrainz, Discogs, theaudiodb, or all three at once. When Jellyfin finds music, it'll scrape all the metadata and create an artist entry, which creates a folder in Jellyfin's data folder. This is all normal. For some reason, sometimes it'll look in the metadata folder and go "Oh hey, there's a folder here," and instead of just reading the information to load the artist, it adds it as a separate entry.

Sometimes removing the offending music, thus removing the artists, then adding it again solves the issue, but then, in the case of Band-Maid, the issue comes back eventually.

[quixotic120]

I have also tried removing the artist, re-adding, and even removing the folder within \metadata\ to no success for certain bands. Interestingly the same as you I am noticing it appears bands with a non alphanumeric character like hyphen, hiragana, or apostrophe in the title are most likely to have issues

[Dex Luther]

I have also tried removing the artist, re-adding, and even removing the folder within \metadata\ to no success for certain bands. Interestingly the same as you I am noticing it appears bands with a non alphanumeric character like hyphen, hiragana, or apostrophe in the title are most likely to have issues

On second thought, after looking at other artists, having the path as

Code:

.\metadata\artists\

appears to be the correct one. The one pulling from the library's content folder appears to be the duplicate and the incorrect one. 

Code:

.\Music\

[quixotic120]

ah interesting, this had me take a closer look. mine is not the exact same as yours but similar.

Scraping with musicbrainz and theaudiodb:

Code:

.\metadata\artists\

results in an entry that has a musicbrainz ID, some albums, and images but no description or link to audiodb (just link to musicbrainz)

Code:

.\Music\

results in an entry that has a musicbrainz and audiodb ID, all albums (including those listed in the above entry), images, and description (a more complete entry