I’ve disabled the linked recording, now there’s only the good one 
(I compared the pattern first)
1 Like
Trying to understand that initial question again.
1. specific recording including correct title - yes, of course
2. correct recording (content of file), but tagged wrongly - I suppose no
3. different but similar recordings (?)
Number 2 you would correct at MusicBrainz and that would then let the data linked to AcoustID be correct.
Number 3 is something we can’t correct from MusicBrainz. All we can do is delete obviously badly linked data.
I think one question you have left is about that compilation. Harder to answer without having a copy of the compilation to test with. It could be the compilation has been made up with tracks from this live CD. That would be a good reason why it appears on this AcoustID page. On that 3 Deuces page I would remove the Japanese link, but leave the compilation unless I can prove otherwise. (i.e. a track list from the compilation that tells me it is not from this gig)
This is also a good way of reassuring you that you are dropping bad data. Between all of the clues I don’t think anything is left to say that this single submission has any quality to it.
1 Like
I’ts possible, that it is the same recording. It is similar in length, but I would compare the pattern of the live album to some other versions of 3 Deuces. Although I don’t think they will match.
There’s not much known about this compilation - Discogs offers microscopic images and a note that #7 (that’s it) is misspelled as “Deuces” - no further information …
1 Like
Whoops - I went back and edited my post. I agree with you this is going to be the same track. My deductions as follows.
It is a sampler CD from the same year. Only a single set of AcoustIDs attached to that Release - usually more reliable. The Discogs art is useless to read at 300dpi, but maybe “The Ozzell Tapes” is written on that cover? I can maybe see that shape in the middle of the second sentence?
I would merge them.
1 Like
I think that’s reasonable - done it.
These are exceptionally small - I can’t read anything from the back.
1 Like
This is the interesting part of digging around in AcoustIDs, it allows us to spot links like that. I have often found unusual compilations and samplers with tracks of an artist I am editing. That then leads me down a rabbit hole of buying that compilation on EBay and coming up with a heap of other related artists to chase. 
With the small art -
Zoom HARD on track seven an use imagination
The Ozell Tapes? Maybe?
2 Likes
If I know what is printed there, I can actually see it 
Your probably right, and in fact it’s not a big surprise
(although I will compare the live and studio recording of 3 Deuces)
EDIT: partial similarities:
overall: no chance to confuse these tracks. That’s maybe possible for pop songs, not for jazz.
This is the trick I often use. The brain knows what to expect to see, so can we trust to read that? the word in front of my yellow highlight could also be “album” and we know then end of those waffle is always the copyright stuff.
It was more the case this is a sampler from the same year as the album was released made me go for the merge. Context fits well.
This is why playing with the AcoustIDs of tracks you own is always a good move. You get to see how different they usually are.
What happens when you slide that downwards? Does the black lump move down the page? This happens with timing errors when one tape is running slower. Edit: Have played with that one now, not enough lining up. Needs to be blacker to say it is a match.
1 Like
No, it vanishes. I suppose it’s connected to the bass line. In the live recording the bass continues to be dominant, in the studio version, trumpet is taking over. But that’s far from certain. But these versions are really different enough - it has to be the live recording… (but F/US or shortened Japanese ???) 
EDIT: Is it possible to see which recording was submitted with which pattern?
I don’t know what you are comparing as all I could do was to hack the URL to include the two fingerprint IDs you showed in your image. If you are comparing studio to live then yes, it makes sense some bit will show similar, and then they will become very divergent.
It is fun comparing old Floyd gigs when someone has recorded it from the radio. You see the timing errors there as a black block will move down the line as you shuffle the offset. Showing that different tape speed but the same source.
This example just didn’t have enough “black”. Not enough to say it is really the same. Or the same with some noise \ mastering.
As to “is it the live recording - F/US or shorter Japanese” can that not just be spotted by length?
If we are back to this one: Track "9d372daa-b3a7-40d2-a1d9-a415a8c97e7c" | AcoustID
Then I would disable the 6:42, and merge the 6:25 and 6:23.
Yes, that’s what I’ve done.
Both the European and the Japanese share the same acoustID, but they have different track length and there’s a second acoustID with a separate pattern for the Japanese. The Sampler version shares also this acoustID, but I don’t know, which pattern was submitted. There are pattern for F/US and JP on the same acoustID. Is it possible to find out?
Would that change anything? I suppose all three pattern would remain… (regarding disable)
I’ve initiated the merge.
They “share” the ID as someone has incorrectly linked that 6:42 recording to the 6:25 AcoustID. As you pointed out, previously there was no Japanese edition so they just picked something that looked close before incorrectly submitting fingerprints.
The only certain way to find out is to get hands on both F/US and Japanese editions. Which one do you own? That would be a start point. Load it up in Picard and Scan it. Then you can look through your acoustIDs and check them off.
From a visual compare, I see the Japanese edition has a longer recording for this track. That 6:42 recording has two AcoustIDs attached. One match the length, the other doesn’t. So I would deduce that the shorter AcoustID is incorrectly linked and would disconnect it…
To make it clear I have now disabled what I believe the the bogus ID. See how now we have recordings that match the AcoustID lengths in each case.
I also did that here: Track "285d7b3b-faeb-42e6-916d-b09b0b756e69" | AcoustID for the other unique Japanese edit
1 Like
I see. Now both Japanese and F/US have only one remaining acoustID, but the one from the F/US release has still 3 pattern… (and all 3 are very similar, one of them has an offset of 4, the other one matches without offset)
By “pattern” do you mean the bit at the top of the AcoustID page? That is the “Fingerprint”. The pretty dots are the audio fingerprints that Picard (and other taggers) submits.
Then AcoustID website does some magic and create the AcoustID number at the top of the page that links (almost) identical fingerprints. This AcoustID is what we then use to match to Recordings in Picard.
As MusicBrainz editors we can unlink recordings, but not remove fingerprints.
(Or that is at least how I understand it…)
1 Like
I mean the fingerprints
Patterns sounds good to me. They looks kind like what happens when you roll a finger over an inkpad 
(Oh, and sorry if my waffle ever repeats something that is obvious to you… sometimes I just waffle to make it clearer to myself while explainin)
1 Like
Another strange thing from the studio album (again 3 Deuces):
Looks like a MBID - is this what remains if recordings are merged?
from: https://acoustid.org/track/98558225-18f7-4f25-b297-fa032c4020a0
I believe if you had a Japanese edition, and a F/US edition of this CD, then you would have two totally separate sets of AcoustIDs. I think the lengths here are showing they both have unique fingerprints and AcoustIDs due to their differing lengths, differing chunks of audience \ fade outs.
MB has merged these recordings as they are the same recordings and the only differences are caused by those bits of extra audience. Or “early fades”.
This now leads to pairs of different AcoustIDs attached to some of the Recordings.
Yes, that is an old merge. Click it and you go to the same 3 Deuces track as listed above it.
I assume you have seen some HUGE messes like this on tracks like Sledgehammer (maybe not - that one is pretty clean).