Spotify knows your next favorite song — here’s how
- The Whirl
- Dec 5, 2019
- 3 min read
by Bea Sancio
I’m not exaggerating when I say that music has taken over my life.
Like almost everyone I know, a huge chunk of my listening is through Spotify, a digital music service that was launched in Sweden in 2008, initially having more losses than profit. By 2011, the company took a turn when it received $100 million investment that was used to launch in the United States, and it’s only been skyrocketing since then. Today, Spotify has over 50 million tracks, 232 million monthly users, and 108 million subscribers (as of June 30, 2019).
Spotify has done a pretty great job at recommending songs that make up the perfect soundtrack for my dull but weird life. Sometimes, though, it can do too good of a job.
Yep, it can get creepy. It’s like Spotify knows my music taste even better than I do.
How does Spotify do it?
Here’s a crash course on how the Spotify algorithm works. In a nutshell, there are 3 types of recommendation models in place.
First of all, Spotify curates music through the collaborative filtering model. What this model does is find similar users, and go with the assumption that they will enjoy each other’s tracks––those they haven’t listened to yet.
There’s also the natural processing model. Spotify takes it to the web, scouring for posts and written texts on different artists and songs, to find out what people are saying about them. Afterwards, Spotify weighs these descriptions to determine if 2 music pieces are similar.
The last one is called the raw audio model. Unlike the first two, this model takes new songs into account, since it focuses primarily on the audio itself to figure out its personality.
Let’s say you like Dancing Queen by Abba. What does its raw sound say about what kind of song it is?
From there, you get recommended songs that form a similar personality, new or old.
Okay, now that you know how Spotify curates our music feed, so what? How does it affect our listening experiences?
Well, I’ve found that the Spotify algorithm hasn’t helped me discover new music at all. Instead, it’s created an echo chamber of my music taste––of what has been tried and tested.
Spotify has banked on these traits of mine––passive with weirdly specific but diverse music choices––to effectively shape my overall listening experience.
Back when Spotify wasn’t a thing, most people used to listen to new music through albums. Nowadays, though, we’ve formed a playlist-centric culture.
This playlist-centric listening culture has also stopped me from fully immersing myself in an artist’s discography.
At the end of the day, Spotify is a service with profitability in mind. This is why it gives you songs that its algorithm knows you’ll like. How else would they keep you subscribed to their Premium services monthly?
If, like me, you actually want to start exploring other kinds of music and artists, solely relying on Spotify’s recommendation models isn’t the way to go.
It’s time to go old school!
Don’t get me wrong, I still love Spotify; it remains to be my favorite application.
It’s totally fine to keep enjoying Spotify and its music recommendations, but we must learn to be critical of its implications and limitations. Ultimately, this technology is used to capitalize on our interests and emotions.
Spotify has undeniably changed the way we listen to music, but maybe it’s about time we transform our listening habits––on our own terms. SOURCES:
Comments