Hi, I’ve been playing with Picard for a few days and have hundreds of unrecognized Clusters (about 30%). I’m just starting to go through these and some have duplicate files which (I assume) prevent the cluster being recognized by MusicBrainz. I dragged one cluster (a double CD) over to the corresponding icon in the right pane and ended up with duplicates there.
I didn’t want to have to look at the info on each track to decide which to Remove so I Removed the CD icon and the Cluster, then deleted the duplicate files. When I add the folder(s) again, nothing apparently happens, with no change in the number of Clusters or Unclustered files. The log just says “
Adding directory: '[path to directory]' ...”. I’ve tried adding the m4a files with Add Files and drag and drop.
How can I fix this?
If I close the window will I have to start again?
What’s the best way to find and deal with duplicates?
I’d appreciate any help.
Windows 10 Picard v 2.7.3
Deleted them from the folder on your hard drive? Then they shouldn’t reappear when you add the folders.
If you’ve deleted them in Picard, if wont delete the actual file. So a re-import would add them again.
With duplicates, you can just hit ‘save’ in picard, and then search your Music folder for ‘(1)’. Check that none of the results are songs that actually have ‘(1)’ in the name… then delete them. It’s not perfect, but if you don’t care about what quality you keep etc it will work fine.
Why your 30% might not be matching is a completely different bunch of paragraphs so let’s put it aside for now
There are a few possibilities here:
- The files are actually still loaded and where not among those you removed. Maybe they are still attached to a track on the right pane somewhere?
- The files get freshly loaded, but already have been saved with MusicBrainz identifiers and hence automatically get matched to the corresponding track in the right pane.
- You directly drop them onto something (an existing cluster, a loaded release) but expect them in unclustered files
- It’s an actual bug where some error occurred when removing the file and Picard still considers the file loaded
In case of 4 there might be directly some error message in Help > View Error/Debug Log, that could help identify the issue. But only a restart will be able to fix the state again.
In all cases it would help if you could enable debug output in the error / debug log and reproduce the loading issue, then share the log here.
Also a general advise: Don’t load thousands of files into Picard at once, work with smaller amounts that you can actually manage.
Thanks, it was number 3, I found about 6 of them in another folder after Expand All. Is there a way to search in either pane?
Will a restart mean I lose my work to date?
I did load the entire folder of recordings when I started in a brief moment of foolish optimism and because I hadn’t discovered the option to select multiple folders and because it’s expected a program and its underlying services should cope with that. I also had the Classical Music plugin which slowed everything down then crashed, which was fun. Still, it’s interesting to see the success rate in total.
I solved it, thanks. Some of the tracks were in the left hand pane in another folder, probably from drag and drop (see my other reply) which prevented the album(s) from being added.
I don’t want to save to my music files yet because I’m not sure yet that what MusicBrainz has is better than what I already have from Windows Media Player and iTunes.
The 30% unrecognized are so far pretty evenly split between duplicate/missing tracks, ambiguous matches and the release apparently not in MusicBrainz, though the results from “Search for similar albums” don’t necessarily find the album even if it is there; “Search for similar tracks” is sometimes better - maybe I should look at the advanced search but that adds more steps and time. Many of the CD’s I’ve loaded to help recognition are not in the MusicBrainz DB and again, loading each CD is a chore even if it is in the DB, never mind adding a new release and that most of them are up in the loft.
Is there a comprehensive workflow with the best and most efficient way to go about this whole process (more than the outlines under WORKFLOW RECOMMENDATIONS on this site)?
What is it you are trying to do? The workflow recommendations cover what I would suggest.
It often good to make peace, from the start, with that fact that there is no ‘magic’ way to easily pull better tags (accurate ones) across a large music library, using Picard or any other app.
If you can clarify what your goals are then we can look at if Picard can meet those expectations
For instance, better tags than iTunes and WMP, easy! Getting 100% of releases matched even when they’re not in the DB… well, there’s nothing to match to, so if you don’t want to add any data (manually to your tags or to the DB) then it wont happen
p.s. yes, if you close Picard you will lose anything not applied to tags/saved. Make a copy of a couple of albums and work with them for now. Ideally you go through, at a leisurely pace, save all the easy matches, and then you can assess what to do with what’s left.
My aims are to better categorize my 1000 ripped CDs, with accurate, authoritative, consistent and more complete information than I have including composers, genres, performers, cover art, inserts and booklets and I hope to do this as efficiently as possible for the 300 remaining.
Then I have to consider comparisons between old and new (why are the tags better here?), saving changes then repeating the process with the Classical Music plugin enabled for the albums in that category, saving a backup each time to be careful.
I’ve looked at the workflow and though helpful it’s limited, for instance the enormous gulf between 3 and 4, which ignores the cases (I have 300) where there is not only one match:
3. Select the correct release
If there is only one release that matches the lookup, it will be loaded automatically. Before proceeding, please check to ensure that it properly matches your album (e.g.: release country, date and label, catalog number, barcode, media type, and cover art). This is especially important if you are going to submit any information such as AcoustID fingerprints.
4. Match the files to the tracks on the release
Drag the files from the left-hand pane and drop them on the release in the right-hand pane. Check that each track on the release is associated with only one file. The release icon should turn gold. See the Matching Files to Tracks section for details.
After my last post I discovered Tools menu → Tags From File Names and have recognized another 11 albums with
and I’m working through other patterns. This is an advantage over working with a few albums at a a time. I don’t want to do this at “a leisurely pace”, that’s what computers and automation are for. I’m trying to see where the flaws are here in the workflow as presented by documentation and the program so I can work round them. That’s what I need to see in a workflow.
If some tags are particularly borked then it can mess with Picard - e.g. Picard expects the Artist to be in the Artist tag, and a Track title to be in the… you get the idea. Looks like tags from file names is a great workaround!
Have you played with the Picard threshold settings yet?
I would try the ‘scan’ button on any remaining tricky ones that you suspect are in the database - but I would definitely do this slowly/a couple of files at a time. Scan will ignore your groupings and existing tags so you will end with a pile of albums on the right, and will have to manually identify the correct one, and then pull the rest of the tracks into that one.
Computers can automate pretty much everything, in line with the mantra of ‘crap in, crap out’
What they can’t do yet is take a library with mixed quality tags and duplicates and pop out a beautiful golden egg with a few clicks
If you’re not manually checking each release for an exact match after they’ve been matched by Picard then you are already foregoing ‘accurate’ information, at the release level. This manual check requires a ‘leisurely pace’ no matter what, unfortunately. That’s not to say that the broader tags wont be accurate, or an improvement, if you skip it, but it is a point where you need to assess a need for speed vs. a need for specificity.
You need to work on max five albums at a time. Otherwise computers do what they are best at and automatically guess and mess up the data. You need to be able to check visually what is found.
And don’t panic. 1000 albums will take a few months, but your methods will get faster and faster checking. Listen to the advice above and you will soon tame this system.
(And make sure you have an original backup before you start so you can refer to it when you get a bad match)