If 2014 was the “Year of the Horse” in the Chinese Zodiac, it was the…
Chuck the Analyst Finds Radiohead’s Most Depressing Song Using Science
It isn’t every day you find out you work with a celebrity—or, in this case, a viral sensation. Charlie “Chuck” Thompson, one of our incredibly talented data scientists, was recently featured on The Verge, Esquire, Gothamist, and others for designing an algorithm that ranks Radiohead songs by sadness. The answer? True Love Waits, with a shockingly depressing Gloom Index of 1 on a scale of 100.
Charlie hasn’t applied his model to AudioBlocks stock music yet, but he does have a lot of interesting ideas about how data and art intersect. We asked him some questions to see what learnings he can offer our creative community:
Audioblocks: Why did you set out to discover Radiohead’s saddest song?
Charlie: I found the Spotify dataset a while ago, and the valence metric (“a measure of musical positiveness,” according to Spotify) really stuck out to me. The idea that an algorithm could tell you how sad a song was fascinating, and I instantly thought of Radiohead since they’re my favorite band.
AB: How did you decide on your method to determine sadness?
Charlie: I initially was just going to use Spotify’s valence metric, but it turned out that two of Radiohead’s songs, “True Love Waits” and “We Suck Young Blood,” both tied for saddest song. I couldn’t accept a tie, so I decided to pull in lyrics. After doing some research on sentiment analysis, it seemed the most intuitive method was to use the percentage of “sad” words per song.
To combine musical and lyrical sadness, I wanted to account for how “important” lyrics were in a given song (A mostly instrumental track, for example, should have less weight given to its lyrics than a rap song). Coincidentally, someone previously analyzed Radiohead lyrics and came up with the concept of “lyrical density”—the number of words per second—so I used that to determine how to balance valence and lyrics in the final equation.
AB: What about data and music inspires you? How do they converge, in your opinion?
Charlie: I love the blank slate you have when either starting a data analysis or writing a song from scratch. While music is more intrinsically creative, the wealth of data out there allows you to explore basically any research question you find interesting. I’ve also always thought music was fairly mathematical, and I think it’s amazing how a bunch of sound waves can affect your mood.
AB: How does it feel to be a viral sensation—and are you a mad data scientist?
Charlie: I never thought there was so much overlap between data nerds and Radiohead fans! It’s been super exciting to share my two favorite things with so many people. I did once dress up as a mad data scientist for Halloween, but day-to-day I try to keep the madness contained.
AB: Of course, we have to ask, what is your favorite rock n’ roll AudioBlocks song?
Charlie: Fried Chicken, for sure:
As the first subscription-based stock media company—including stock footage, images, and audio—in the industry, VideoBlocks (the AudioBlocks parent company) knows what matters most—it’s people. And we’re prouder than ever to brag about the success of our fellow Blockhead, Chuck!
Want to be a part of our close-knit and talented team? You’re in luck—we’re hiring! And be sure to check out more of top rock n’ roll tracks—depressing and upbeat moods included.
|Cheer Up with Stock Audio|