Hello! I need to add Genre and maybe some other tags into my existing library.
I have tried Picard with some albums but it did not matched albums correctly, creating 1 album with track 14 and other almost identical album with rest of tracks and many similiar troubles.
I have all music correctly sorted this way “Artist/Album(year)/01. Track.flac”
All tracks are correct in their folders. Can I make picard consider folder structure in matching process?
Yeah, when you have that large of a library, you don’t want to make the same mistake as I did at the beginning with using “Scan” (a useful, but dangerous tool). When you add your tracks into Picard, if they don’t have any MBID’s in their tags, they should appear in the left pane. Once all are added (all tracks I’d imagine would be listed under Unclustered), select all the tracks (CTRL+A) and click Cluster from the menu bar. This will then group them based on the album folder. You should be able to right click on each cluster select ‘Lookup Similar Albums’ (I believe that’s what it’s called). Picard should hopefully have the correct album at the top (as long as it’s in MB). Be sure that genres are enabled in Picard Options and configured as you want them.
Happy tagging!
Thanks, but that’s still around 78000 albums, even if, it’s a lot work correcting all of them. Not to mention picard is having trouble loading them all at once. I had to stop it because it was freezed for 2 hours. I need it to follow folder structure as primary source of “metadata” for matching because that’s 100% correct and it should work without any manual work.
You say each folder is an album? If the files also contain a corresponding album tag then the clustering should organize the files nicely, and lookup will be able to.perform album lookup.
If the existing album tag is not useful or consistent, you might want to use the Tags From Filenames function to generate an album tag based on the folder name, then cluster the files.
Well, don’t do this. Work in batches. You were also planning to check the results before saving, right?
See Picard just stopped working — MusicBrainz Picard v2.12.3 documentation
As said, you really want to work in batches, especially if your collection is not well tagged.
Since it is ordered by artist/album, I guess you can start with one artist with many albums, or few artists (starting by A).
10 albums at once will let you review & correct, and it is actually faster.
Once your collection is tagged, updating is much faster (since mbid are embedded in tags), but 78000 albums is still a lot… Think that every album loaded use resources (memory/network/disk IO) and depending on your options will lead to one to a lot of requests (which are rate-limited).
As well as working in batches, a HUGE tip. Make a backup of your current collection before you start. Partly so you can unwind any errors. Also it is possible you will learn more about Picard in the process and may way to redo things a bit later.
Also go into the options and tweak the match settings a bit.
Bias it towards the types of releases you have, countries, etc.
I have tried it on small batch outside of library. problem is that it incorrectly match tracks from one artist. instead of let’s say, deluxe album, it match 1 song from standard album, show it as incomplete and rest of songs are in deluxe as incomplete because that 1 track is in standard… and so on… there’s many errors like this.
I think that most basic tags is OK, like artist, album, year and track name and it’s number, but it somewhat get it wrong anyways.
You have given a good example of why it needs manually checking. Picard can’t tell if that track is in the normal album or the deluxe. So makes an educated guess. You can then manually drag the track back to the matching album.
Another item to be aware of. Once a album has been chosen on the right, you can Right Click the album and select “Other Versions” to swap to a potentially better match.
Working in batches of a few artists a day will not take you as long as you think. It’s only safe way to not make a huge mess of a library.
yes but all tracks are in one album folder with year, so it should not split it into 2 albums (standard, deluxe…). There’s exact number of tracks that matches standard album, they are numbered… I guess picard see every track as standalone and read tags of that track but it do not care about folder structure which may fix all issues with wrong match in this situation.
clustering should stop any splitting? otherwise yes Picard will just match files to whatever it thinks is best
I didn’t think clustering stops splitting. I see the same issue myself at times. Though I am so used to just reassembling and shuffling release versions I don’t really pay attention to any pattern.
after I cluster them, it is in left pane, looking almost good except some unclustered tracks (dunno why). after clicking “lookup” it will match albums and move them into right pane and sometimes it move track between albums making duplicate in that one album, with missing tracks in another (correct) album, or other similiar errors.
It shouldn’t do this if you lookup the cluster instead of individual files. Lookup on a cluster searches for matching releases, this will never find multiple releases.
Also use lookup, not scan. Scan always works on a file by file basis and matches by audio fingerprint.
I open folder containing all albums, then select “unclustered”, run cluster, then select clusters and tap on lookup. correct? or should I go album by album from cluster list?
Yes, that’s correct.
If you get multiple results then for an album this would mean clustering already created more than one cluster for an album, which indicates the album tag is not consistent across all files
well, I am tired already with just single artist after 30 minutes… it’s unable to match half of the albums(singles or similiar) even manualy. Can I somewhat just force it to write Genre to tracks and not overwrite anything else? that way, I would be at least partialy fine for me…
There’s issue with finding albums… Does MusicBrainz know all albums and singles? If yes, I need to re-check everything
Assuming that a single FLAC file is 50MB on average, you have 200,000 songs in your 10TB library.
I only have 14,000 songs on a 64GB flash drive.
Do not press the Submit AcoustIDs button, as this will clog up the already large fingerprint queue.
no - if its not in the database then it won’t be found.
Can you give some examples where something isn’t being matched and we can then confirm if the failure is because of missing data or for some other reason