YouTube Users Are Reinventing The Way We Classify Music

It's not just about 'Pop' and 'Rock' anymore; now we've got 'Hair Dryer Sound' and 'Music For Babies'

YouTube users have helped define “crowd-generated music categories.” — Illustration: Diana Quach
May 23, 2016 at 12:53 PM ET

The power to watch endless music videos on YouTube and stream virtually anything you want on Spotify has changed a lot of things about the way we listen to music. But it’s also changing the way we label music. Gone are the generic “rock” and “soul” record store bins of days past, but what’s left in their place? By analyzing the songs that pop up in the “related videos” bar of YouTube, a group of academics discovered that the way music is lumped together and consumed on YouTube doesn’t correspond with traditionally-defined genres at all.

The study—conducted by Massimo Airoldi, Davide Beraldo, and Alessandro Gandini, and appearing in the journal “Poetics”—showed that in addition to the genres we know and love, there were a host of more novel categories, like “Hair Dryer Sound” and “Glee Music.” These genres often contain videos for which the “situational purpose” is emphasized over production qualities, like “Music for Babies” and “Relaxing Background Music.” Another interesting discovery about these new categories was that the word “hour(s)” frequently popped up in the title, revealing that how long a song or audio clip lasted was just as (or even more) important than who was performing it or whether it was of good quality.

Without access to the proprietary algorithm that dictates which videos are served up, the study’s authors had to back in to the process. To do so, they pulled a sample of 333 music videos, and then pulled 25 of YouTube’s related videos for each one. After weeding out any non-music content, they were left with 22,141 videos to analyze. A content analysis of the videos’ titles revealed which words most frequently recurred in each cluster. Often these corresponded to traditional music genre classifications, like hip hop, pop hits, and country. But the surprise was that some of them did not. For instance, good luck finding the “Guitar Tutorial/for Musicians” section of your local music store.

To get further insight into the relationships within each cluster of videos, they tagged each of 10 most common words in each video’s title. The tags they used included “milieu” (the country the music comes from), reception (words like “sleep” and “relax), “genre,” and “cross-genre” (words like “live” and “remix”). They also accounted for words that related to the music business (like labels, radio shows, and venues) and the platform of artists (YouTube stars, Spotify, etc.).  Ultimately, here’s what they determined:

“Our analysis shows that the retrieved music clusters are not meaningless agglomerations of videos. On the contrary, there seems to exist an underlying cultural logic of similarity within each cluster that is produced by the technologically-mediated and aggregated practices of usages by listeners and uploaders.”

Long story short, the means by which we seek out and and consume music on the internet is changing the way we relate to it.

Some of their findings go against what we think of as sacrosanct to listeners: who is the artist and what type of music is it. Of course, this does still factor in for a lot of the clusters. But this analysis reveals that beyond grouping music according to its various artistic merits, people will also relate songs to one another based on functional or practical reasons—like will it make their baby stop crying.

This behavior is also evident on streaming services like Spotify, where playlists exist for all kinds of moods and situations, like “focus” and “travel.” But even beyond the habits of listeners, you can see this same concept playing out amongst those who create music. If you go far enough down the rabbit hole that is Soundcloud, you’ll not only find music that sonically melts the barriers between genres, but comes with whole host of new category names to reflect this ambiguity. You can find anything from the hyper-minimal +, to nightcore, to trapwave, to soundclown, to the oft-debated vaporwave. Often the songs in these categories share little common sonically, but are connected by other means—region, inspirations, techniques, etc. All the while, any preconceived notions about what these categories mean will mutate until they eventually splinter into new microcategories. (Hard vapour, anyone? An eccojam, perhaps?)

Taken as a whole, this gives us an insight into what the study’s authors call “crowd-generated music categories.” Because the study was not using a representative sample of the entirety of YouTube, it can’t deliver a definitive list of all the genres which appear on the site, but it does give us a starting point. And more importantly, it shows us that how we talk about music can change based on how we’re interacting with it. Rather than imposing a top-down fixed list of musical genres, we can instead see it as a universe in constant motion, shaped by each selection we make.