by Carolyn Heneghan
We know apps can play, stream, store, share, identify, and even recommend music—but could they also be responsible for predicting the next hottest songs, artists, and albums?
Earlier this year, Billboard announced its Streaming Songs charts, which take into consideration the top picks from services like YouTube, Vevo, Spotify, Muve Music, Rhapsody, Xbox Music and more. These music apps play a part in delivering information about the current top spots for songs and artists. But instead of simply proving which are global favorites based on plays, purchases, and other aspects of big music data, some music apps are beginning to use that information to predict future favorites.
Shazam claims they can. The song identification service just announced its Top 10 Breaking Artists of 2014 based on “user curiosity,” and Billboard seems to back its viability. But how can Shazam be so sure of its predictions?
When Music App Predictions Have Hit the Mark
But let’s back up a bit. Before jumping into next year, we can look as far back as 2011 to see where some of Shazam’s music predictions did actually hit the mark. In short, Shazam identifies music based on short clips of the song it hears through your Shazam-enabled device, so the app’s biggest pool of data comes from the songs identified the most. They use this song tag data to formulate predictions for the next hottest song, artist, or trend.
One of Shazam’s earliest predictions was that hip-hop would dominate 2011. While Top-10 lists can be extremely subjective, MTV’s list does in fact confirm this prediction, with half of the top 10 being hip hop artists. Hip hop also stood for about 30 percent of Billboard’s Top 50 songs of 2011. While hip-hop didn’t sweep the Grammy’s, Shazam did hit on the right notes, so to speak.
In 2012, Shazam released another set of predictions, this time for the summer jams of 2012. At the end of May, the service predicted that, based on song tags from its users, Gotye, Usher, Maroon 5 ft. Wiz Khalifa, Calvin Harris ft Ne-Yo, Rick Ross ft. Usher and Nicki Minaj would be among the top summer artists for 2012. Again, while top music lists are entirely subjective, many of these hits appeared on many lists released throughout the rest of the summer season.
With 2013 came both a more definitive judgment of these big data predictions plus a little friendly competition. Just before the big awards show, both Shazam and music streaming service Spotify released their Grammy predictions based on their own user-generated data sets. The results were surprisingly close and surprisingly similar.
With correct guesses for four out of six categories, including Album of the Year, Record of the Year, Best Country Song and Best Pop Duo/Group Performance, Spotify made out with 67 percent accuracy. Shazam went for even more predictions, and having correctly surmised 11 out of 16 categories, its accuracy percentage won out at 69 percent.
What about SoundHound’s accurate predictions for some of the 2011 Grammy nominees? Or Pandora’s ability to predict more music you’d like based on hundreds of musical attributes for each song? Shazam isn’t the first to harness big data in light of the future of music, and it certainly won’t be the last.
What This Means for the Music Industry
Armed with endless data points, Shazam and other music apps have found a way to become a bionic ear to the ground for the music industry. While not perfect—and what prediction could be in light of the subjectivity of music—these predictions are often eerily accurate. And what’s more eerie, perhaps, is that organic human tendencies and preferences can be so astutely broken down and processed by big data and computers.
Imagine where the music industry would head if some of these apps combined their sets of data points. What if Pandora could take Shazam’s predictions for the top new artists of 2014 and then analyze each artist’s musical attributes to come up with another set of even farther-reaching possibilities for top breaking artists of next year?
To take it in another direction, what if music industry leaders, such as record labels and producers, take Pandora’s analyses of those top artists’ attributes and create songs that embody those very qualities—turning those predictions into the next manufactured set of top-40 hits? Could boiling down music into tiny bits of data end up killing the creation process and churn out the music that a set of programs tells music listeners they want to hear? If we’re afraid that much of top-40 is already manufactured as it is, what kind of harm could ultra-specific data sets do to rig the sways of popular opinion regarding music as we know it?
It isn’t all doom and gloom—when it comes down to it, no one is ever going to squash musical intellect and innovation. But it begs the consideration of what it means for music when it becomes more digitized and thus more removed from its original organic art form.
While big data has granted these apps the ability to make informed predictions about the future of music, it hasn’t enabled them to be right all the time (see Grammy predictions 2013). However, as big data gets bigger and pieces of data get more refined, you may see these predictions becoming even more scarily accurate over time.
Comments are closed.