Workflow Question

This is probably one of those “you’re doing it wrong” issues but nonetheless, here it goes:

Even at the very beginning of the clean-up and tag process, my folder structure is already pretty proper, meaning Artist, Album, Year are all already in the folder name. Usually even in the preferred order Artist Year Album.

So, after playing with Picard for a while, here’s a few questions:

When I drag a bunch of those folders over to the cluster panel, why does everything go to “unclustered” first, when in fact everything is already properly clustered.
I then have to click once to move it to “clustered” and once more to “look up” and move things over to the right hand panel where it then starts to look up things.
This seems an awful lot of unnecessary clicking for a rather simple task.
I mean no offense at all, just trying to understand why I seem to be unable to get this done with let’s say one click.

Then, when I drag to folders over with the same album (but not the same release) at the same time, the above process gives me one album in the right hand panel with all the songs lumped in it.
When I do the same thing but one album at a time, each gets recognized properly and shows properly right hand.

Frank Sinatra - Christmas Dreaming (1987, Columbia)
Frank Sinatra - Christmas Dreaming (2019) (24-96)

Those 2 done separately, each gets recognized properly. The first as the 1987 CD release and the 2nd as the 2019 remaster of the 1957 release.

Those 2 dragged over at the same time, then clustered, then looked up gives me 1 cluster with 21 files and looked up 1 album with 11/11/21.

How can I drag over a large amount of albums and get them treated each separately so the various versions/releases get recognized as such?.

Just figured out where this goes wrong.

Picard seems to use the current meta data to lookup (naturally and logically lol) but since this is exactly the data I want to fix, it’s not really useful.

Any way I can get Picard to ignore (or use it with lesser priority) the current meta data and use the folder name to cluster and lookup?

I just retagged those test files and once Sinatra albums differ (e.g. one says Christmas Dreaming 24/96) I can drag them over together and they get clustered separately and then looked up as separate albums.

There is an open ticket about allowing automatic clustering. Because of the way file loading works this has been a bit tricky to implement. So currently you need to do 2 clicks instead of 1. I wouldn’t call one click “an awful lot”, though :man_shrugging::slight_smile:

When deciding which release matches best Picard gives a slight preference to already loaded releases, to reduce spreading separate files over multiple releases.

Use “Tags from file names”, see Generating tags from file names — MusicBrainz Picard v2.6.3 documentation


I didnt say that. I said “…a lot of unnecessary…” but mea culpa, I tend to be much more demanding with things that are almost perfect, like Picard. The mediocre stuff ain’t worth nit-picking at. :grinning:

I already do this with Mp3tag. No worries then, I was thinking, if there’s already a way for Picard to ignore meta and just use the folder names supplied, I could save another step in the whole process.

I cannot speak for @outsidecontext, but personally I would prefer that people do ask sensible questions about potential improvements that they would like to see. Auto-cluster is one of them.

Another might be whether the Tags from Filenames function could be improved or made more automatic.



Most of the questions that arise while I’m using Picard irritate me. Because I think, it’s just plain logical that everyone would want e.g. auto-cluster (or no cluster at all, just use folder names and go straight one-click look-up).

But then again, I havent been around long and what I consider logical or an optimal work-flow might not be suitable to everyone or even most.

Tags from filenames is another one that irritates me. Why go that complicated route? Why read the folder names, then use them to convert to tags, tag the file, refresh and see how Picard now works the way I would have suspected it does anyway, use the initially loaded folder name to go straight to (unclustered, clustered, looked-up) voilà.

I do understand however, having been with the likes of SickChill, Radar and such for a long time, stuff grows organically and at some point it might no make sense anymore, even though it did, very much so, not such a long time ago.

@loungebob It would be helpful if you would focus your comments on explaining how things could be made better not simply what you don’t like about things.

So for example, how would you like Tags from filenames to work, bearing in mind that your naming structure might be:
Artist surname, initial/Album title/CD 1/1. Track artist - Track name
and someone else’s might be:
Album title - Album artist name/1-1. Track name - Track Artist

So for example how could Picard know that it was Track name - Track Artist and not Track artist - track name? Or is one format so MUCH more common that it can be used as a default?

Pretty sure this is considered constructive criticism since I supplied how I think it would be better.
I am saying, if I already have proper folder structure and naming (but the files inside arent yet tagged properly) why not tell Picard to use the folder names to cluster/lookup instead of going the tags from file names.
If I can tell Picard my folder naming convention is e.g. Artist Year Album it could use folder names as is and wouldnt have to do any additional steps.

Use it for what exactly? People have organized their files in a variety of ways, and actually for many they use Picard to create an order where there was non previously. The folder could be an artist name, and album name, a genre or any combination of those, or something completely random. Tags from filenames is flexible enough to handle this.

In you case, if I understood correctly, the tags are unreliable but the folders are named very consistently. But that’s your music collection, not everyones.

Picard actually will guess artist and album name from folders if the tags are empty, but it prefers tags if they are set.

But honestly in all the years of Picard’s existence only very few people ever asked about the tags from filenames functionality. And automatic clustering was asked for, but turned out to be tricky to implement. And since that it is literally only a single click the importance if this small improvement was obviously not considered that important by most.

It is a pretty common fallacy to assume because one needs a specific feature that everyone else does as well.


It’s a good thing then I literally said that my way might not be suitable for everyone or even most.

Never mind, I’ll do this part in mp3tag.

1 Like

Just to ensure I did not come over as dismissive: you’re requests are of course totally valid, and we even have an open ticket about clustering by directory only instead of by tags, and we definitely should do this. But indeed this is not something that was asked for a lot, especially compared to some other still open things.

For most open suggestion the answer to “why isn’t this being done” is usually “because nobody did”. As most open source projects Picard lacks developers and time, and the wishlist is long :slight_smile:


As I said before, it is always valid to question why things are done in a certain way and whether they could be improved.

However, I disagree with Philipp on the reasons “why isn’t this being done”. Here is my list of reasons:

  1. Too few people are asking for the feature
  2. There is no consensus on how the feature should work - different people want it to work in different ways
  3. The new feature would conflict with or break existing functionality - or the tradeoffs in e.g. usability would be bad
  4. The new feature would be difficult to implement or the performance would be pants.
  5. The new feature requires functionality from elsewhere (e.g. MusicBrainz) that does not exist
  6. The new feature doesn’t have the interest of a developer
  7. The new feature doesn’t really belong - Picard has a niche, a purpose, to tag files using MusicBrainz data. Adding a coffee making feature might be something thousands of people will like, but…
  8. It is something that could be achieved by a plugin.

What I read; coffee making could be achieved by a plugin. And I like it! :joy:


That is not so far from the truth. A plugin could be used to instruct your smart switch to turn on the coffee maker.

But no one is ever going to agree to Picard core code doing that. :slight_smile:


If Picard can make my coffee as well then it really will be the do it all tool. That sounds like adding perfection! :coffee: