Organise your music with Picard
Organising your digital music collection can be a Herculean effort. Yet when you have several gigabytes of tracks to sift through, your only chance of finding what you want to hear is if your music files are properly tagged.
Don't panic if your machine can't tell Barry White apart from the White Stripes, though: the creators of Picard feel your pain. That's why, by the time Picard is finished with your music collection, each file will know the album it belongs to, the artist who performed it, its track number within the album and a host of other details.
Picard is simple to get hold of - it's available in the software repositories of most distributions, or you can download Picard straight from the MusicBrainz site. That means it's just a matter of running either the apt-get install picard or yum install picard command to get started. However, while the default view when you launch Picard is sufficient, we think that it adds an unnecessary degree of separation between you and your files. Thankfully, Picard also includes a far superior File Browser view, which you can switch to by going to View > File Browser.
Now that you've changed the view, any new files you add to Picard will initially show up in the middle panel under the heading Unmatched Files. Click on one of these files and the left-hand pane of the bottom panel will display any metadata currently associated with that file.
Our first step in getting your music organised is to cluster all files from the same album under one heading. So, select the files you've just added and click the Cluster button on the toolbar. Depending on the album metadata present for each of the files, Picard will group all songs from the same album together. Files that don't belong in a clustered album will remain listed as Unmatched Files.
To get more information about your music files, select songs or a cluster from the Unmatched Files list and click the Lookup button in the toolbar at the top of the interface. Be aware that Picard may take some time to retrieve information about the selected songs, depending on the speed of your internet connection. If the program can't find the data automatically, you can also find track information manually by clicking the Lookup button in the New Metadata window. This will summon your browser and point you in the direction of the MusicBrainz database, as shown in the box below.
Step by step: Finding metadata manually
Search the web: If the automatic search is proving too cumbersome, use the Lookup button in New Metadata instead. While the Lookup icon in the toolbar searches using information available in the songs, the button on the bottom-right searches using the information provided in the New Metadata column. You must provide an Artist to proceed.
Choose your album: If you haven't provided Picard with an artist's name, you'll be prompted to do so now. Browse the available options and when you've decided on an album that seems appropriate, click the green Tagger button next to it. You can also use the Tag Lookup section at the bottom of the page to further refine your search if necessary.
Matching up the data: Back in the Picard interface, the album data should be presented in the far-right panel. Now drag the song from the Unmatched Files and drop it on the album name, not within the album. The song's counterpart in the album will light up if it's a match. Click Save in the toolbar, or right-click the song and click Save to store the new metadata.
After fetching the relevant information, Picard displays the name of the album that the song or songs belong to in the panel on the right. Picard also removes these songs from the Unmatched Files list and moves them into this new album entry in the far right panel. Double-click this to expand the album and view the new metadata for the tracks.
Now it's worth noting that Picard uses each track's previous metadata, if any, to make intelligent guesses at pairing the correct data with your tracks. It will also show you how certain it is of those guesses by displaying a small coloured rectangle next to each track - green is a good match, yellow, orange and red represent increasing degrees of uncertainty.
Inspect each track in turn, right-clicking the song and clicking Save if the software has got it right. This tells Picard to attach the new metadata to this song. As a visual indicator, the coloured rectangle changes to a green tick to show the track information has been saved.
A green rectangle next to a track represents a good match between current and retrieved metadata - check yellow, orange or red tagged tracks closely.
Dealing with false positives
For songs that are matched incorrectly, you may find they've been confused with another track in the same album. In this case, rearrange the tracks into the correct order by dragging and dropping them and then click Save. If the data is completely wrong, however, you'll need to right-click the song or group of songs and click Remove.
Note that you'll have to drag and drop these tracks back into the middle panel before you try to retrieve any new information for them, because removing false positives in this way won't automatically reassign the tracks to the Unmatched Files list.
With that done, you can try adding some information to the song yourself (the artist's name, track length or album name are helpful). After that, click the Lookup icon in the toolbar at the top again and Picard is more likely to find an appropriate match.
Fingerprints for easy identification
Picard also enables you to use audio fingerprinting (akin to taking an audio snapshot of each track) to create a temporary key that can retrieve track data. This works in collaboration with MusicIP, which operates the MusicDNS service, comparing fingerprints to Portable Unique Identifiers (PUIDs). However, the procedure relies upon a closed source, patented technology called Music Analysis to create new PUIDs, which - depending on your ethics - may well put you off.
You see, while you may not have to use any proprietary tools yourself, this entire process depends on some other user having previously performed Music Analysis on the same track. It's also why Picard can't perform Music Analysis directly, because the proprietary tools used for it can't be integrated into an application under the GPL.
If you're happy to go ahead, click the Scan button in the toolbar after you've selected a song in the Unmatched Files list. Picard will then calculate a fingerprint of the selected song and query the MusicDNS servers. If a match exists, Picard receives the corresponding PUID and then looks through the MusicBrainz servers for a matching track.
Pick your MusicBrainz
Driven by a community of music aficionados, MusicBrainz is an open alternative to the CD database. In other words, MusicBrainz collects user-contributed metadata information, such as title, artist and album, then makes it available to the public. Picard is an official MusicBrainz product, but it isn't the only application that employs the free repository of music metadata.
Some of the other Linux applications with MusicBrainz support are Amarok, Banshee and SoundJuicer. Most of the data on MusicBrainz is in the public domain, with some content available under the Open Audio Licence.
First published in Linux Format magazine