I have been working on a side project: clean up years-worth of accumulated digital music. Some was downloaded. Some was ripped from CDs. Some was ripped from vinyl. One thing you notice after accessing enough media: the metadata isn’t always great. Even when it comes from a ripped disc, you get the sense that music publishers aren’t spending a lot of time on metadata. Here are some of the tools I’ve been using to get things straightened out.

Hands On Metadata

When Windows 10 came out, Windows Media Player disappeared for a bit in favor of Groove. It reappeared after a bit (or I learned how to resurface it) which was handy. But if you right click on an album in Windows Media Player, the resulting search was not always able to find a match. So to start, I did a bit of the work by hand.

You can actually do a lot of metadata management at the file level. Drop into a folder of MP3s – this will not work with all sound files – and select an albums-worth to edit their metadata. Right click, choose properties, and select the Details tab. Now you can change the fields (like Album name) that apply to all the files.

If you don’t see options for artists, title, album name, etc., you may have a mixture of file types. Selecting multiple files and entering properties only shows you the common fields. I sometimes get this when I accidentally select an album cover (.jpg) with music (.mp3).

This led to some interesting discussions with our kids, who are very into music and have opinions about its storage. At one end of the spectrum, a person who puts all of the music files into a single folder. The music is retrieved and played using playlists – album or original container agnostic. At the other, a person who puts things in folders and worries about both the filename (invisible to the music player) and the metadata (which is visible).

Any foldering system immediately creates decision points. Are folders by performer? That works with some genres (pop) better than others (classical). Or else your “various artists” grouping expands. Who is most important? Composer? Conductor? What about “my greatest song featuring Lil’ Me” or a violin soloist on a larger recording?

I get both. The ease of dumping all music into a single folder and ignoring foldering in lieu of metadata. And the need to organize for retrieval outside a music player. When you rip or download music, you tend to start in a folder-ized world. That is what we had but the folders were not consistent (artist, creator, composer, whatever). So one of the things I started to do was to choose (artist) and then winnow down the other folders to true up the rest.

Add Some Sophistication

There are some nice tools out there, though. Someone put me onto the free Music Bee player (also in Windows Store), which does a great job of helping with album art and tagging. It is also an easy way to fix small things with metadata. For example, say you have a 2 disc set (often titled by the music publisher as Title Disc 1 and Title Disc 2, not just Title), it is easy to denote that they are related as discs. You can do this in the file properties too (look for Part of Set and add 1/1 or 1/2 or 2/3 or 3/3 or whatever) but it’s handy to have this in-player. Something simple like this can collapse a multi-disc set into a single group in a player.

One benefit of using a music player is that you can see metadata issues visually. I can open an album and see duplicate copies. If the music lives in different folders, I may not have noticed the dupe at the folder level.

Another thing you can do with a player is grab album art. In a lot of the music I download, the art is already embedded. But if a music player does not read that, then you get no album art. You can right click on an album in Music Bee and choose Edit. When you right click on the Album Art, you can choose a picture or have it find a match.

The matching isn’t great. I found that you have to have the Album Artist, not just Artist, field populated for a good match. Also, unless you want to automate the album art selection – which I don’t, because it’s not 100% reliable – you may not get the prompt to save a copy as folder.jpg or cover.jpg, two of the common album art file names. In this case, I found I had to repeat the album art search (or reapply a picture I had already downloaded) to trigger the options to save a copy.

Copy Catalog

Sometimes the metadata doesn’t seem quite right. For example, I ripped a vinyl copy of One Step Beyond by Madness. The album cover lists only 14 tracks but there are 15. One site that I am using – and have loaded my own collection into – is Discogs. It is not comprehensive but it has a lot of content, particularly older vinyl. I was able to confirm what the listing should be and that it wasn’t just that I’d chopped a track in half.

Also, while the quality of input varies, these are people who think like library catalogers. Anything that can be documented is and what can be linked connects to supporting pages. For the most part, Discogs is something you look at and re-enter into your own metadata. It is also a great source of album art.

Another useful tool – with as many idiosyncracies as Discogs – is MusicBrainz, which bills itself as the Open Music Encyclopedia. It is user-created and vast. Unlike Discogs, though, the music in MusicBrainz has been fingerprinted. If you use the MusicBrainz Picard tool (available in the Windows Store), you can fix missing and broken metadata.

Picard requires a bit more care than MusicBee. You select an album folder in Picard and it will try to find matches in its fingerprints. Sometimes it’s a direct match and you can grab a ton of really good metadata (supporting artists, publisher or label, etc.) that might not be on your own rip or download.

Sometimes not so much. If it can’t find your tracks, you can select to cluster the files. Here’s what the interface looks like. I have loaded two sets of files. One is for an album – Skarmageddon 3 – that Picard found. All the items that match are in green (on the right side) and when you save the metadata, it applies a check mark.

On the left, you can see a list of names in white. They were not matched in Picard so I have clicked the Cluster button on the button bar to put them into a cluster. Now I can check to see if they are listed in MusicBrainz. Right click on the cluster folder (Swann: Songs in this case) and choose Lookup in Browser. This will execute a search in your web browser on the MusicBrainz site.

I have not found this first step to ever hit the right album. In this case, the ripped CD had bad metadata. So searching MusicBrainz for the album using bad information won’t retrieve good information. Instead, I searched for the composer (Donald Swann) and was fortunate to find the album listed under its actual name (Songs of Donald Swann).

Another way is to right-click on the cluster and choose Generate AcoustID Fingerprints. Then right-click and choose Lookup – not Lookup in Browser – and see if it will make a match based on the fingerprints of your files. This also worked for this particular album. Once found, I could save the enhanced metadata over to my local files.

Localize It

Once you have grabbed enhanced metadata, you may find that you are finished. As soon as I saved the Songs of Donald Swann metadata, the album was renumbered and appropriately split in two. Once I refreshed MusicBee, the album appeared under So- instead of Sw-.

You can obviously still add local touches. One thing that I am doing is using the MusicBee album art feature to download larger album art. Much of the art that was ripped in the 1990s with Windows Media Player is 200px square. An album art search in MusicBee will return images from Bing and iTunes that are 1200px square. I am trying to grab larger files to get the clearest album art I can.

I am trying not to rename albums but will occasionally add a tweak. We have George Michael’s faith, both the original LP and the later 2 CD box set. Sometimes I have an LP and then also had a CD of the same album. In most cases, I will note in the comments if I ripped a digital file from vinyl. But sometimes I’ll also note the difference in the title name. A CD version may have had more tracks but the same album name.

Some music players lump albums with the same titles (“Greatest Hits” for example) into a single group. Adding my own tweaks can make these easier to separate. One area I have not worked out are the symphonies. Lots of albums entitled Symphony this and Symphony that. There has to be a better way to wrangle those to make them easier to identify. I’m thinking that the solution is search (composer name or something).

The number of tools available is amazing. Having worked in a library, I was not surprised that there was no single source of truth when I started looking for music metadata. I have done some original cataloging (such as it is) on Discogs for albums I own that weren’t already in the database. One wonders why the music labels don’t upload their own historic metadata to help with this (one wonders if they’ve been asked, or if they have it). But it is good to see so much crowdsourced activity around capturing music and information around music.