Rather than relying on musical components or a binary rating system, a new website crowdsources playlists based on emotions, moods and activities.
The key to a great playlist isn't a rogue's gallery of artists or a carefully balanced mix of genres. Great playlists are inherently emotional, the soundtrack to our lives. They can pull at your heartstrings, create much-needed moments of zen and turn your darkest moment into the best day of your life.
That's the concept behind Stereomood, a relatively new music site taking an innovative approach to online recommendations. Part Internet radio, part discovery engine and part mood ring, the music aggregator builds on Pandora's much-praised model of utilizing listener feedback, but with a twist: Rather than tagging tracks based on specific musical components or evaluating their relevance through a binary thumbs up/thumbs down system, Stereomood relies on users assigning a "mood" to each song. The engine then groups songs together under relevant playlists, with moods ranging from the straight-forward ("calm," "meloncholy," "dreamy") to the situational ("just woke up," "busy as a bee," "dinner with friends" and the ever relevant "afrodesiac.")
"Behind every song there's always an emotion. We don't know why but maybe that's why we love music," Giovanni Ferron, a senior web designer at Xing and the mind behind the online radio, wrote on Stereomood's website. "So we've created a way to suggest songs that follow your feelings: Stereomood is the emotional Internet radio, providing music that best suits your mood and your activities. How do I feel? What am I doing now?"
Conceptualized by Ferron in 2008 and brought to life with the help of a dedicated group of friends, the Rome-based music engine relies on an active and engaged user base to continually tag and adjust Stereomood's evolving taxonomy of mood playlists. Each playlist is generated starting with the user tags; tracks are arranged based on "an average between the tag numbers and the date of issue of track itself." If users disagree on tagging -- a listener declares "Incinerate" by Sonic Youth to be 'Uplifting' rather than 'Depressing' -- they can add a brand new tag, increasing the depth of Stereomood's playlists. The results is an intricate web of songs, all clustered along Stereomood's constantly growing taxonomy of moods, emotions and activities.
Stereomood's sourcing goes beyond just users who visit the site. Stereomood brings in new tracks from a selection of the best international music blogs: "If a post contains an MP3 link, it adds those links to our database and display them on the frontpage," Ferron explains on the site. "Each related post can be reach simply by clicking on the 'read full post' link below the track description." The system allows Stereomood to constantly pull new music from audiophiles around the globe; customized buttons on certain partner sites allow listeners to tag tracks in the Stereomood database without leaving the comfort of their favorite blog. Stereomood's blog network is fairly extensive, including American indie favorites like Pitchfork and Stereogum alongside Mexico City's 8106 and Kalender 08, a free vintage music blog in Sweden.
The mood-based criteria for Stereomood's tagging system hold some fascinating implications for the future of crowdsourced music engines. The past year has seen several advances in semantic-based analysis of the social space, including the much-trumpeted use of Twitter to predict changes in the stock market based on a "social mood" extrapolated from the vast sea of tweets sent on a daily basis. Why not apply the idea of a mood-based recommendation engine to the broader social sphere, rather than simply rely on active users to shape Stereomood's body of playlists? There already exist services like Social Mention that scrape mentions of a product or brand from Facebook, Twitter, YouTube and Digg and, through semantic analysis, determine how people are discussing a certain topic (if only in the sense of "good" or "bad.") I could eventually imagine Stereomood generating playlists perfectly attuned to chatter of music fans in the Twittersphere.
Past forays into social music have been less than successful (iTunes Ping, anyone?), and while Pandora remains the archetypal recommendation engine, its insular model of relying on a single user's preference makes it a less than perfect model for a database built on collective action. Until then, Stereomood's mood-based taxonomy and open tagging model remains an interesting, social alternative to thinking about music recommendations on the Web.