You have a dream. You want to be a rock star – a rock star with a hit, maybe even more than a one hit wonder. What could your jam session in the garage possibly have to do with data science? It turns out that big data can help to predict hits based on some specific musical characteristics. According to the folks at scoreahit.com (who created the “hit potential equation”), certain features of a song -- such as loudness, tempo, danceability and energy level help them determine whether a song hits or misses. Crazy, huh?
These researchers from University of Bristol’s Intelligent Systems Laboratory have studied the UK pop charts from the last 50 years and are able to predict a hit with a 90% accuracy rate if they adjust for changing musical tastes in each decade. (Apparently, it is harder to predict hits in more creative and innovative periods of musical history such as the early 80’s – stop laughing, you know you loved “Tainted Love” — than it is in other decades where hits are more consistently similar – hello, 1970’s.)
Meanwhile, we ponder the following: will the ability to predict hits lead to what some of us would argue is the further homogenization of popular music? Will, or should we say, are songs being written specifically to become chart toppers? (Ahem, Rihanna and One Direction.) What does that mean for the creatives, the artists, the innovators? Will they have to write to the equation or just work Youtube to hit it big “unexpectedly”?
Author: Tracy Menasco