Was playing an old Macca album yesterday (Pipes of peace) and ‘Hey Hey’ was reported as ‘Hey Jude’. Inspecting the listen shows :
"track_name": "Hey Hey"
but:
"recording_name": "Hey Jude",
File is 100% correctly tagged - checked with MP3Tag, Audioshell and Media tag (and even opened it in Notepad++)
Clearly a bug/error - can/do these things get fed back as bugs?
Welcome at the forum @trgz
Is it possible the track does not exist in the MusicBrainz database? If it isn’t there, LB cannot identify it.
If so, the remedy is simple enough: add “Pipes of Peace” to the database and link the track.
@biocv The rest of the tracks on the album were correctly titled - could it be that the database entry for that one specific track was incorrectly entered originally?
Update: I’ve been through a few (most/all?) of the CD releases of the album in MusicBrainz and that track is always correctly listed as ‘Hey Hey’.
Update2: I found that I could search the Recording MB ID (1e668b1b-09be-4f72-baa1-2fd6a233157e) for that track, and it shows the wrong album ('Maida Vale Studios 2013), and that tallies with the ID for the album (26d9a9a4-a459-490c-bc4d-57c1a14f3bbf) in the data (though not an album I own, or have ever played), yet, as mentioned, the rest of the album appears to be correct.
FYI most of the other tracks show ebe26f4d-27cf-46ef-8474-4f9a8d284c3b as the album ID, though I notice that there are two others that do not have any album art, and may not have been correctly identified.
Something that I didn’t mention was that this is being scrobbled by SimpleScrobbler - could the disconnect be there? It’s the first time I’ve seen anything in a very large number of scrobbles (though another may have slipped through)
Hmmm, that is a bit odd. If the track exists in the database it is usually found (though not always the correct version). Perhaps there is a typo somewhere that threw the mapper off?
Anyway, did you link the correct track? That should stop this particular mismatch from happening again at least.
I have to admit that I haven’t a clue as to how to go about doing that and, given what I’ve found, my gut feeling is that I wouldn’t really be fixing the problem.
Welcome, and a big thanks for taking the time to report 
Can you link to the specific listen, or the page with that listen on your profile? Then I will tag in some of the LB code wizards to have a look.
@DontMindMe posted a lovely explanation recently. It is quite easy.
And yes, that will fix your problem in the sense that all historical and future listens of that particular track will be mapped correctly.
Could you say a little more about that? Presumably, that is only guaranteed for a given player/scrobbler, since other scrobblers could extract and submit a slightly different subset of the metadata?
Does this apply to listens reported by others too? What happens if two people submit inconsistent mappings?
The nitty gritty details are better answered by actual members of the LB team, like @rob or @lucifer. What I know is what I pick up here at the forum.
As I understand it, listens are linked to the database using hash values based on the the artist name and track name. So as long as these two bits of information are identical, the listens will be linked to the same MBID.
If two users manually make the same link, it becomes available to other users. I do not know what happens with conflicting data.
Thanks!
Two tracks can have those pieces of information identical. For example, https://beta.musicbrainz.org/release/6f4f77b8-a948-4b53-9e86-7f8f15a64384/disc/1#55609c58-9420-3b53-b973-6e5846952137 and
https://beta.musicbrainz.org/release/dd21b3ee-c206-4e7e-8f7e-aab0208152b2/disc/1#3b2a618c-e67c-4d50-bc19-4e241f576880 both have artist name “Fritz Kreisler” and track name “Liebesleid”.
That isn’t an obscure edge case, it is quite common in the music I listen to.
Presumably, the team is aware of this possibility. I’ll check out the links you shared.
That is another aspect which can be very cumbersome indeed. I understood your first comment to be about differences in the data of the same audio file which various scrobblers submit to LB. How LB goes about picking the right MBID from various identically named recordings in the database is another can of worms. The problems you mention for different performances of the same classical piece also plague live performances of popsongs. A similar problem that I often encounter is that LB maps recordings of dance classics to DJ-mixed versions in the database.
If only all scrobblers submitted the MBID, this could be solved.
You read my original comment correctly. Your reply prompted me to think more about my recent frustrations with unmatched listens.
Yes, it would be nice if more players reported MBIDs. Even the official Android listenbrainz player doesn’t report MBIDs when I play tagged mp3s (Android API and/or permissions issue?).
There are alternatives though…
Listens from Spotify and YouTube come with their own identifiers, which could be matched against the MB database. I’m tempted to write a script for that if I work out how to search MB by 3rd party identifier and submit mappings of listens to MBID rather than messybrainz ID to MBID.
I have created an issue for this: https://tickets.metabrainz.org/browse/LB-1593