Metadata management in Audio Files

The starting point for adding metadata to a track on a cd is the CDDB data but the source of that data has been something of a lottery. While the original FreeDB is no longer available, it had been copied via various forks and a commercial branch, and while there is a more direct path to gnuDG, the alternate MusicBrainz is a much better alternative. The major change is that this iteration goes a lot further than simply providing disk and track metadata for the cddb. It provides a much more complete set of metadata but the cddb set is still rather restricted although some of it can be included in the extended data field as long as it is restricted to 256 characters.

In order to access the additional metadata, MusicBrainz provides a number of other services and these are used by their own graphic tool, Picard. What I had not appreciated initially was just what Picard was providing, and that is adding metadata directly into the ripped audio files. The are still a few little niggles that have to be taken care of and one of the more problematic is the use of colon's in the metadata which is then used to create the file names for the tracks. Not a problem in Linux, but a major blocker in Windows which neither FAT or NTFS allows. I have had a few other problems with metadata used in the cddb file which then prevent the transfer to a memory stick or windows formated disk as used with SmartTV's and the like. I had initially thought this was that double quote was a problem, but I now think that this may be something that only becomes a problem when there is a single one which then messes up applications that wrap the text in double quotes in order to use it. I've ended up manually switching the colon's to semi-colons in The Classic Composers set of disks in order to archive them on a memory stick. At the moment, the internal metedata in the files still has the colon's in it.

One thing I had not appreciated was that the editing I was doing in Picard was for the metadata set that is attached to the file, and it was nice to find out that once the audio files have had this data added, the management tool I use to check between mirrors, BeyondCompare, will decode and display the differences between files. This was one of the reasons I had started using BeyondCompare in the first place, but for encoded development files. Metadata is a bonus.

The major addition made by MusicBrainz is the management of all the credit information and then in the on-line views these are all cross-linked to pages combining all the albums that a composer, performer or publisher that have so far been uploaded. I am getting a lot more confident on uploading a new disk to the database. The core elements are quite easy, but adding some of the relationships can be a bit tricky. This is not helped when the entity you are trying to link to has a foreign title. Something I have commented on in edits on-line. That the Russian Composers all have their names in Russian, making them difficult to find initially. That one has to manually add the correct English translation as the credit information shows just takes time, when my own systems have had translations built into the process, showing entries in the preferred language.

While my original crib sheet has a quick guide to adding a new disk, I need to document some of the finer detail, such as how to credit a choir, and highlight the bits which do not seem to be supported, such as 'Artistic Director'. Areas that I have yet to find the prefered way of adding.