bkdelong

So I've gotten into podcasting hardcore this past week, even generating some ideas for my own show. But one of the major things that's been annoying me is the complete lack of proper ID3 name tag metadata in the MP3s that make up these podcasts. There's a reason for metadata - if done properly and even verbosely, search engines will someday be able to search the metadata INSIDE the MP3 file itself just like it will be able to search JPGs, PNGs, and TIFFs for Adobe's RDF-based XMP metadata.

Adam - DSC-2005-03-14 means little to me or my Tivo. Call it "Daily Source Code podcast for 03/14/2005 - 'basic title here'". And don't put the show summary in the "Album" ID3 tag - that should be reserved for the Comments field. You have one MP3 per show so considering a single show an album is silly. I'd call your album "Daily Source Code". Make use of the Track Number and Disc Numbers. Say you have one Disc per month and one Track per day. So this episode would be Track 14 of 31, (provided you do an episode every day this month), and Disc 3 of 12 in the Year 2005. It's like tracking magazine volumes and issues.

And so many of you podcasters end up doing transcripts for your shows with hypertext. I'd say at a minimum you should put the final URL for the transcript in the comments field - if it fits. Perhaps someday the community will make it much easier to redirect "expired" URLs or to have MP3s "check-in" with their originator if requested so they can change metadata from time to time.

Best practice should be to copy the whole darn transcript into an appropriate ID3v2 tag so it travels with the file. Then eventually those who are deaf could read the transcript and those of us who are archivist metadata freaks can keep the textual data with its original audio. And, of course, searching for any text in the transcript would bring up the MP3 file. The only problem is I'm not sure of an ID3v2 tag that supports massive amount of text. Perhaps making use of the Lyrics3 spec needs to be an option. Basically the transcripts are "lyrics" for a podcast.

Someone high up, (Adam, Ron, Dave), should consider posting these or similar ideas as "Best Practices" or even "Podcasting Recommendations". Because while metadata can't be exploited to its fullest now, it certainly will be in the future. The DSpace archive, for instance, is meant to archive data for over 100 years. The better we tag now, despite the fact of the hundreds of thousands of terrabytes of data that may accumulate, if metadata resides in the same file its describing...it will be that much easier to find, recognize and reuse.

[BrainStream]
(Permanent link to this entry)