Maybe I am just not understanding how Picard works, in reading through stuff it makes it seem like I have to put a folder in at a time and what for it to check it? Maybe I am to tired, or maybe it’s the pain meds ( I had surgery on Friday), but I am not seeing how I set Picard up to just scan my music files. If I have to do a folder at a time, it would take me many months if not years to get through all my music. Any advice or directions in helping me understand what I clearly don’t would be much appreciated.
Depends on the state of your current tags. You can’t just assume Picard will work like magic on a pile of music files. So you need to check the results. Ideally backup the music first. Then drop music in in batches. If you have fairly decent tags already, then matching up releases will be easier. If you are starting from a chaotic heap then you’ll have more work to do in sifting out errors.
You’ll need to start slow to get the hang of what Picard can and can’t do, and then you’ll speed up.
It also depends on what your final aim is. Are you trying to get perfect matches to your ripped CDs? Or just getting decent MBID tags for a media player to work with?
1 Like
I had been using Windows Media Player, so all my files were/are tagged with that format and I would like to keep it that way.
What happened is our (business cloud) NAS box had a drive go bad and when it went bad it corrupted a bunch of the files on the other drive (which isn’t supposed to happen) by which I mean, it mislabeled tons of songs. Right after this happened Windows stopped supporting media players connections to musicmatch-xboxlive so I have had no way to scan the drive for errors and have them fixed. I’m a DJ and when someone requests a song and put in Hot & Cold by Katy Perry and get California Girls by David Lee Roth, it’s not a good look for me. So when I say Extremely Large Database I mean millions of songs ( somewhere between 500 gb and 1 tb). So appreciate any and all advice and help I can get with this.
Ouch. You poor thing. Windows Media Player. Such horrific artwork… 
But this at least is album\artist\track names. So a good start. So lookups should do you well.
In Picard you likely want to bias the matches. Options \ Metadata to - i.e. country, more album\single than compilations.
But where is your music from? Did you rip this as albums? Or heaps of tracks from soulseek\torrents?
This is why you need to manually check stuff as you go along. Don’t assume a computer can give a perfect result. It may get 90%… but that is 10% shock Katy Perry tracks.
Not that large, but it takes time and patience to get correct. You are the human who can give a final check. Only you will know what is good. So if you want to avoid giving Katy Perry fans David Lee Roth then you need to look at the results as you go along.
I would start by a backup. (Something you have clearly learned about since the NAS disaster) And put that backup away in a drawer. Now have a “unsorted” heap. And a “Sorted” heap. Move files in batches from one to the other. Start slow. Start with stuff with the best tags as you work along. And double check a good sample.
As a DJ your best bet is focus on a small subset of your collection. Get that correct first. Then widen out your work.
As DJ I assume you are coming from real CDs and not just a heap of junk from pirated file sharing. So I’d hope your collection already has a structure.
Don’t assume you can dump 1TB of music on Picard in one go and get perfect tags. Not a chance. You’d end up with a few too many Davie Lee Perry matches… Quality takes patience.
I would ask what is the target here? Just clean tags for a media centre? OR perfection matching exact releases?
1 Like
Ok, thank you for the advice, I will give it my best shot. And I thought the NAS was my back up, but now I have a multi-drive NAS instead of just a dual drive. And I have a back up drive that sync ever 3 months..
As to the music it is all either burned from CD’s or bought as digital downloads from trusted sites, nothing is burned or ripped from a torrent site. We have been in business way to long to risk losing it over not being legitimate.
As you are coming from properly organised files, this gives you a good chance to get good matches.
Do look at the Options\Metadata\Preferred Releases and tweak a bit. Example:
If Picard finds multiple albums that look the same that match, this makes it lean in a direction. Example above I am more likely to match a UK CD than a US Vinyl. More likely to match full albums than compilations.
Other tips from my own use:
-
Drop some albums on the left side. Hit “Cluster” to group them by their Album tags. Next hit “Lookup” to lookup matches in the database based on the tags. As the matches happen album moves to the right to check before you hit “Save” to write tags.
-
When checking your results sometimes you may find an album split in half between two versions. If that happens you can drag tracks around on the right hand side to bring back as one album. (Can happen with some “deluxe” or “bonus tracks” albums)
-
Sometimes on the right side you get a match of the wrong version of an album. You may hold a USA CD and it matches a German one instead. Right Click the album and you’ll find on the right click menu “Other Versions” that lets you swap around between editions.
-
If you fail to get a good match, then “Lookup CD…” can read a rip log and look for an album match that way.
-
Using “Lookup in Browser” will take you online to the database. You can more directly look around at alternate matches. Watch for the little green buttons. These little green “TAGGER” buttons on the page will then let you copy that album back to Picard.
-
With good basic tags you should not need “Scan”. This option uses an audio signature. More like “listening” to a track. Trouble is it may match into compilations and other sources of the tracks a bit too often. It will find the recording, but may not be your exact one.
-
If you are after artwork, then consider doing all your matching first without artwork. This will work much faster. If you are skipping artwork on your first run to focus on the metadata matches you’ll gain in lookup speed. If you then return later with artwork enabled this will be quicker as the second time you drop the album into Picard you’ll not need “Lookup” as now it has the “MusicBrainz Release ID” and instantly knows what album it is. Doing an “artwork only” pass can then be done in much larger bulk, and you can just load up hundreds of albums and walk away while it is downloading as you won’t need to be checking the matches this time.
-
Be warned that some popular albums can have dozens of versions released over the decades. Reissues, different factory pressings, minor differences. You’ll need to decide early on as to how fussed you are about exact matches of your copy. Do you need a factory for factory match? Or just good clean tags?
There are plenty of other threads around with similar tips. Start slow with some obvious albums. Learn how to check as you then speed up. I personally don’t use the file renaming and moving, but some do. I work from know folder an artist at a time. In batches. And the more I learn, the more artists I can drop in at the same time.
You WILL mess up on the first few attempts… and then eventually get the hang of it.
Just a note about Picard, the version 3 will be released soon, and it contains a bunch of improvements that should make working with huge set of files a bit easier:
- matching algorithm was improved, Picard 3 will find more often than Picard 2
- memory use was reduced, especially when loading a lot of tracks/albums
- performance when handling a lot of files was improved
- it has a way to load/save sessions
When it comes to tagging, there’s no magical tool, it all depends on the existing tags, the existence of the tracks/albums in the MusicBrainz database and/or in Acoustid database.
Since you have music mainly from CDs or legal digital downloads, results should be good, be sure to follow recommendations about the workflow:
- always Cluster first,
- then Lookup
- if album isn’t found, you can look for it directly on MB, and if missing, add it. You can copy an url to search box in Picard to load any album
- you can move files to tracks manually
- Scan only if needed (Acoustid lookups, slow, it requires proper configuration)
- Sometimes it finds a release, but not exactly yours, use context menu on the found release to switch to another one in the same release group
- Do it in multiple passes: disable cover art and extra relationships on first pass, so it doesn’t downloaded all images, or make extra queries once properly tagged you can do a second pass and add missing metadata.
- Always store MBIDs in files (that’s the default), but also use MBIDs on load (that’s also the default)
- work on a small local set (copy files locally, tag them properly, move them back once clean/well tagged), working on network is always slower, and sometimes unreliable.
- a small set is something like ~1000 files / ~ 100 albums, try to have complete albums if possible (if album has 9 tracks, be sure to associate the 9 matching files)
- always work on copies (it will ease starting over until you have your Picard settings right)
- ask help here in doubt
3 Likes
Good info from @zas, just keep in mind that second pass will overwrite any localy changed tags unless you exclude them (either manually on case by case basis or better, by adding persistent field info into tags).
While scan can be slow, it finds stuff like nothing else, and you can start it before going to sleep or equivalent.