MusicBrainz replacing correct tags with wrong ones?

musicbrainz
Tags: #<Tag:0x00007fe3d3deb700>

#1

Im still new to Picard and have been trying to use it to get my library of 40000-60000 titles cleaned up. So far I wasn’t really successful at doing so.

My biggest issue is that MusicBrainz keeps replacing correct tags with wrong tags & splitting complete albums into multiple incomplete ones.

For instance:
My album “The Art of Motion” by Andy McKee was previously correctly tagged and in a folder-structure indincating that all the songs belong together. MusicBrainz splits off a single track into a new album with only “yellow accuracy” replacing its correct tags with wrong ones and thus creating 2 incomplete albums instead of having 1 completed album. You can find a screenshot of this here: https://imgur.com/a/0zJgLEW (along with a few more examples similar to this).

Is there any way to prevent this from occuring or -at the very least- reducing the odds of it happening?
As I said, I’m fairly new to this so it’s quite possible my settings are messed up.

Thanks for your help =)


#2

Picard can’t be perfect. So you will always have to act as the human with the brain checking what the automatic machine takes a “best guess” at.

Personally I always work in batches of very small numbers of albums and double check everything.

Your screen shots are good examples of Picard being almost right, but you then need to drag a few tracks around into the the correct album.

This is probably happening as MusicBrainz does not have an AcousticID for every track. Or sometimes you may have the exact same tracks with same AcousticIDs appearing on multiple releases (Deluxe editions especially).

So it is very common to get results like you have in your example. Just drag and drop those tracks matched to the wrong album up into the correct album before you hit Save.


#3

One important advise: When you already have clustered our files as a first step, always start your matching attempt using the “Lookup” button. Given the files where correctly clustered this will keep clusters together. Only if this fails use the “Scan” button on the files, and with this you have to be more careful and check the result.

It’s because while scan will usually reliable the recording (unless there is a mistake in the AcoustId database), it will only look for the audio and not for existing metadata. That means while it identifies the recording correctly, it might mismatch the release the recording is on.


#4

I’m not expecting it to be perfect, I’m just wondering why it gets these “simple” tasks wrong, dismissing correct information and every indication that the songs should be put into a single album and putting songs in different albums.
In all these example I only tried to ID a single album, too so that can’t be the issue.

Overall I feel like MusicBrainz should strive to maximise the number of tracks that end up in completed albums. That should ultimatly reduce the amount of wrong tags to a minimum (at least in my cases).


#5

That’s actually great advice, I’ve only used the Acoustic ID function thinking it would be more reliable =)
Thanks


#6

The trouble is there are often a dozen different versions of exactly the same release. Different countries, deluxe version, remastered re-releases.

Have you used the “look up in browser” button yet? It can be surprising how many different releases there are of a simple album.

Also, like anything clever, Picard can be a bit awkward and stubborn. Soooo many options. But after a while you learn to tame it and get used to how much power it really has.

Give it time. Some trial and error. And you’ll find you soon tame Picard.