The future of music is metadata.

Sonically speaking, each song has a fingerprint. Information-wise, there are any number of parameters to reference: release date, BPM, chart position, etc.

But so little of music’s power is about what we hear or what we think about it. What matters most is how we feel. In 2008, we still have woefully inadequate metadata for the emotion of music.

Genre classification systems were always a blunt object with which to delineate certain emotional themes and forms in music. With the advent of the Web, genres were further broken down into a cascade of sub-genres, as the sheer volume of musical information dwarfed a mere warehouse full of CDs.

These days, niches are getting more ‘micro’ and indie bands how thrive in a direct fan-to-artist industry economy with only a few thousand albums sold. Because the Web knows no limit to the quantity of music available, we’re gonna have to get this thing organized.

It’s already a mess: just look at MySpace Music. How can you find anything in there unless you stumble upon it yourself?

What would be some parameters that would help sort this all out?

My #1 argument for years has been to introduce a new classification system based around an artist’s ’scene’. This may seem like an editorial designation, or something akin to serious musical taxonomy. AllMusicGuide’s genres are so accurate as to approach a living history of ’scenes’, but it lacks the element that would make a scene-based categorization work: a social network.

If you want to save MySpace Music from being the next MP3.com, start organizing it around music scenes. Cross-reference artists with similar top friends, in similar geographical locations, with similar-sized fan bases, playing the same venues together, etc. Have a street team interface where fans can get exclusive access to a band’s ’secret MySpace’ (like ’secret shows’) in exchange for being a super-fan and street promoter. Go nuts! Just do something!

This is not to say there aren’t great services right now using metadata to drive music experience, but most of them are recommendation engines, driven by fancy algorithm or friend network. Nevertheless, the opinion of the individual is a key piece of data for the modern listener.

iLike and Last.fm have great music discovery via social networking, but friends have been telling friends about new music for decades — the only thing new is the technology they use to do it (and, in the case of iLike, bringing artists and fans closer).

Pandora is the best example of the hybridization of human input and relational database processing. Witness the seamless magic with which it picks your tunes based on an elegant behind-the-scenes system that tracks who likes what. As highly that I prize it, Pandora is a one-trick pony.

Back in meatspace, we’re still using critical reviews and chart position to classify music. More caveman tools. Online, strides have been made. Sites like GarageBand.com sharpen these tools and hand them to a large community of peers, and the end result is that top-charting artists often go on to have at lease moderate success. Other companies like Pump Audio and Taxi.com combine this technology with the opinions of expert listeners to move higher quality music into the media stream.

One of the most exciting projects out there is Hit Song Science, which uses audio fingerprinting and genius algorithms to chart sales and airplay against the sonic character of your song. Just plug in a track and you’ll know right away if it’s going to be a blockbuster or a bust. Their next-generation algorithms (currently under development) go further by measuring ‘innovation/prediction cycles in musical structure.’ Cool.

Music. There’s so much of it being made these days, and the hits just keep on coming.

I started this post by saying that metadata was the future of music and that it needed to address the emotion of music. So after all these advances in metadata, why so serious?

The answer is: it’s complicated. Insanely complicated. Like 100 billion+ neurons in the human brain complicated. Like if you want to know more read this 500-page
‘Music and Emotion’ college textbook like I tried to do. It’s unbelievably academic but occasionally exciting.

People are thinking about it. There’s this crazy project called XPod, which tracks biometrics to collect “human emotion and activity information from the user, and explore how this information could be used to improve the user experience with mobile music players.” Another author imagines the device as like having HAL as your DJ.

Bottom line: the more music we have, the more metadata we need to make sense of it. The best metadata is based on the human reaction to music, and the more ways we have to measure that reaction, the better our browsing will be in the future. In 2008, when everyone is their own DJ, everything revolves around the music discovery process.