Music’s Ongoing Metadata Problem. Please Fix It.
If you’ve ever worked with music tags in a program like iTunes, you’re familiar with metadata. It’s all the information about a song beyond the title and the artist: album title, year of release, composers, genre, publishing information, the track’s ISRC number and so on. This kind of information is ultra-important in the digital era because clean, accurate data ensures that the right people get paid for their music. Bad and incomplete data leads to the kind of you-didn’t-pay-me class action suits now being faced by Spotify.
I’m frustrated by poor metadata all the time. Record labels send me new releases directly through a secure digital distribution system called DMDS and it’s appalling the number of tracks that come directly from the labels with incomplete metadata. Misspelled song titles and artist names. No track numbers. One particular major label never includes the name of the album on songs it sends me, so these files get lumped into the “Unknown” album folder in iTunes. That’s nuts.
It gets weirder the deeper you go into this issue. Think about weird spellings (hip hop, I’m looking at you), punctuation issues and the subjective nature of lumping any song into a genre. And the bigger the database, the worse the problems get.
This is why we need something like the Metadata Summit. From Hypebot:
The fourth annual Summit will feature a keynote presentation from SOCAN’s Michael McCarty, Chief Membership & Business Development Officer, and Jeff King, Chief Operating Officer, who will discuss global revenue, discovery, and attribution, as well as sessions on international infrastructure, new business models tied to metadata, and global publishing data. Featured will be top executives from 7digital, AFEM, Beggars Group, BMI, Consolidated Independent, County Analytics, Dart Music, DDEX, Digimarc, Loudr, MovieLabs, Music Week, PledgeMusic, PRS for Music, Rough Trade Music Publishing, SOCAN, Syntax Creative, and TGiT.
“Ever since we launched the Metadata Summit four years ago, we have been impressed by the number of people who have come out to learn more about and contribute their expertise,” said Bill Wilson, Vice President of Digital Strategy and Business Development for Music Biz. “Metadata is the bedrock on which the digital music business is built, and many of the issues facing the business around payment, discovery, and attribution all rely on metadata.”
First held at Music Biz 2013 in Los Angeles, the event has drawn record crowds in previous years, leading to the creation of the Music Metadata Style Guide, currently in its third update, which provides a set of common naming conventions and data entry standards for companies to use when listing, ingesting, and managing digital music.
Fix it. Just fix it.
I’ve spent more hours of my life then I care to think about on curating a massive collection of digital music, mainly dealing with bad metadata. My preferences on naming have changed over time too, so I revise occasionally, but I definitely wish it was easier as most of it is manual.
I use a program called MP3Tag (which is a godsend! check it out and support the independent creator!) and it includes some automation and scripting abilities, but it’s a bit limited in that not a lot is built-in, and I’m very limited in ability in that area. Also not everything can be automated, given the weird idiosyncrasies of naming, etc that you mentioned.
I wish I could afford to hire a professional to script some of this for me! The big guys could afford to do this though… especially since in the digital era, it’s name once, copy endlessly. Get it together, people!
This is a great step though, these consortiums and whatnot, but I worry that they have these ‘summits’ and discuss a bunch of things, without it actually changing anything (Sort of like environmental summits, UN summits, etc!). I hope that’s not the case though.
I agree overall with your sentiment Alan. I feel that Metadata is still poorly understood and implemented. There are also inconsistencies in its use. I’m not a fan of the genre tag just because some music crosses multiple genres and it’s not fair to choose just one. I rip almost all my media files and use tagging software to tag things properly however I do omit a lot of info. I do this to ensure accuracy and consistency in my library. I have even gone so far as to correct other’s tags.
Music Brainz best practices have existed for a long time, since before streaming went supernova. But every label uses some poor intern to send the files to streaming services and they turn data into shit. I can “see” in my Last FM stats the moment i started streaming because of the tags turning into shit: misspelled album titles, no standardization, the annoying “feat.” bullshit, “remastered version” crap and other stuff that frankly i think is there to obfuscate royalty payments.
As frustrating as it was sometimes, the “wiki” approach to tags that Last FM used to have was self correcting and the best i’ve seen so far.