Cluster by Folder

Picard is great, it finds about 90% of my albums no problem. My question is for the other 10%. I have about 1,000 albums.

Is there a way for Picard to auto cluster by folder?

I know you can manually pull each folder into Picard one by one ,but with hundreds of albums that can be tedious and hard to remember which albums didn’t match. Finding each cluster in groups of thousands is also nearly impossible. Lots of people organize their albums by folder. It would be great to have that option so Picard can find the correct albums more easily.

1 Like

have you tried the cluster buton it will try and cluster you songs by albums i fined it doues an ok job at it then any it missed for an album add it to it by dragging it to the album. if you can fined infomaton on an album i.e the song names who it is by and the album name and any other info like barcodes you can add it to music brains sever then you would Picard to find more

I mean, ideally you would add the remaining 10% to the MusicBrainz database so Picard can also find those :slight_smile: :

Otherwise, as @st3v3p says, Picard has a “cluster” button/functionality. It’s accessible both as a button in the menu by default, from the menu as Functions → Cluster, or via the keyboard shortcut Ctrl+U.


Throwing thousands of folders at Picard without manually checking them will lead to lots of mistakes. Picard is not a miracle worker and can often misplace a track - especially something that is on a lot of compilation albums.

It may be “tedious” to go through hundreds of albums, but it is the only way to be sure that Picard gets the correct matches for you. Throw small batches at Picard and you’ll soon get the tagging sorted. It may take a few weeks, but it is better than the chaos you will get if you just throw thousands of files at Picard and hope. :slight_smile:

Sounds like I need to request the Option of auto cluster by folder. I think it would really help a lot of people.

Or can someone point me to what logic Picard uses to cluster albums. Maybe I can use another program to get my album tags close enough for Picard to figure it out.

I’m pretty sure it’s just the artist and album tags. You should be able to use mp3tag to fill those in if the folders are named consistently.

1 Like

While the actual implementation is quite complex, it basically boils down to comparing the album tag and either albumartist or, if that’s not set, artist. The album and artist names don’t need to be exactly the same, instead they are compared for similarity with a certain threshold.

If the %album% tag is not set Picard will attempt to get both album and artist from the file paths. It assumes a folder hierarchy of %albumartist%/%album% for this.

So if you want to get proper clusters you need to make sure to have album and albumartist / artist set for each file or there is no album tag and your folder hierarchy works.

If tags are set, but are so inconsistent that clustering does not produce good results, Picard does not support to forcefully cluster by folder structure directly. A workaround is to use the “Tags From File Names” function (available in the main menu under tools).

In my example I have files without any tags in a folder structure like %albumartist%/%album% and use Tags From File Names to fill the tags:


After I have applied this clustering provides proper results:


EDIT: There is also a ticket about this


In this example, you can see where the album name has the track name in it as well.

I guess I could manually hardcode a tagger script right then to change that album name field… but if there were a way to have a script prompt for input. LOL.

I suppose forcing a cluster would solve that too, with a lot less code changing.

Hmmm… could that be done with a plug in like how Tags from Filenames works, let one specify a name that would then cluster those things below in a folder named as such.

Oh, wait. D’oh! I just answered my own query. Select them all and change the value of album, then re-cluster.

Overthinking … can run you right past the obvious.
As you were. :wink: