Music Metadata: Making my Music Library
All digital music formats support embedded metadata—information about the music, such as artist and song title. In FLAC files, they are called comments. In MP3 files, they are called tags. When your music player flashes a song title on the screen, it's probably reading it from the metadata.
When you rip a song from a CD you get the data (the music itself), but not much metadata. The only metadata you get is just that which can be determined from the song itself, such as song length. The computer cannot obtain the other metadata you'd want, such as song title, from the CD.
A brief digressionary rant: This is one reason why the CD is a dying format. The music industry could have taken some simple steps that would have made compact disc—the most robust audio storage mechanism ever invented—a valuable archive medium for digital music. Instead, everything they've done has made the CD less valuable and less useful.
Five years ago, the music industry should have started putting data tracks on all CDs with the following things:
- Track and album information in some format such as XML.
- Cover artwork, at a couple of different sizes, in JPEG format.
- A PDF version of the liner notes.
Instead, the sort of enhancements we get are spyware-like programs that try to control copying and executable programs that do little and only on specific chosen platforms.
This means that if you want metadata in your music files, you need to turn to a third-party source. There are a couple of free services available to do that. I've been using FreeDB. There also is MusicBrainz.
Both of these services depend on user-provided data, so the quality of the information is variable. The information often needs correction, but it's better than starting from scratch. This allows me to mark every song track with:
- Album Name
- Artist Name
- Song Name
- Album Year
- Track Number
I also save genre information (Rock, Pop, Blues, etc.), but because that too is user supplied, the genre tags are inconsistent and often wrong. (Classification is hard, irrespective of the bleatings of the keyword folksonomy supporters.)
So armed with a good CD ripping program and a third-party metadata source, I'm ready to build my library. Or so I thought. I mentioned the quality of user-contributed data issue. While that's not fun, it's easy to fix: just review and correct the data before saving. I encountered some tougher metadata problems. I'll talk about one of them, and how I fixed it, in tomorrow's entry.