Before I take you through the world of Genres, there's one hypothesis I would like to put forth, not just about Spotify, but about most other music services, especially when it comes to Indian music. My hypothesis: they’re built for the Western music industry, which just doesn’t translate well to Indian Cinema and its diverse music structure. Let me explain.
First up is Artists.
In the Western music industry, artists or bands are usually the ones to get credited, while composers and lyricists often get left out of the limelight. We don’t know who the composer for Beyonce’s Crazy in Love or Coldplay’s Yellow is. Was it them or did they have help? This would also be the same for indie artists from India. Who has composed and written Shaan’s Tanha Dil, or Prateek Kuhad’s Tum Jab Paas? No one knows from Spotify’s data on artists. On all these songs, only the artist or the band is listed.
Crazy in Love was written by Beyonce, Jay-Z, Rich Harrison, and Eugene Record. It was composed by Beyonce, and Rich Harrison.
Yellow was written by Coldplay, and composed by Coldplay and Ken Nelson.
Tanha Dil was composed by Ram Sampath, and was written by Shaan.
Tum Jab Paas was written and composed by Prateek Kuhad
This works well so far (kind of). When you introduce music from Indian Cinema, it starts going haywire. In Indian cinema, each song has a composer(s), lyricist(s), and singer(s). Now on Spotify, things get messy on how they credit artists. Music directors sometimes show up on the credits but sometimes don’t. Same with lyricists. Singers are ALWAYS credited. For example, A.R. Rahman appears as an artist on Saathiya but not on Dil Se, even though he composed both. For Saathiya, A.R. Rahman is credited as an artist on every single song along with the singers. For Dil Se however, he is credited only on one song that he’s sung. Similarly, for Aiye Meherban, R.D. Burman is not listed as an artist, but for Dum Maro Dum, his name appears in the artist billing.
This inconsistency is frustrating with lyricists as well — Gulzar isn’t credited at all on a lot of tracks he wrote, but he’s credited on the entirety of Saathiya. Meanwhile, Apple Music does a better job by crediting Rahman at the album level and listing singers individually, while YouTube Music credits "Various Artists" at an album level but acknowledges the singers for each track.
Now let’s talk Genres.
IT GETS WORSE, and hilarious, but mostly just worse. Spotify tags genres at the artist level, not the song level. Let this sink in for a minute...
THIS. MAKES. NO. SENSE. It works fine for Western artists who tend to stick to a particular genre, but it just about holds true for them. There’s nuance lost there too. For example, Taylor Swift is tagged as “Pop”. That’s it. So all her earlier songs that should be classified as Country or Folk Music, would still be considered as Pop. Bruno Mars is tagged as “Pop” and “Dance Pop”. WUT? Finesse, 24K Magic, Uptown Funk all lose their strong Disco, Reggae, Funk, R&B influences and are lumped in as Pop and Dance Pop instead. Uptown Funk literally has the word funk in its name.
This gets exacerbated in Indian cinema. A.R. Rahman and Shankar-Ehsaan-Loy are both charted under “Filmi” and “Modern Bollywood,” regardless of the diverse genres in their work. Imagine Saadda Haq and Jashn-e-Bahara being tagged the same way - it's music blasphemy — and I don’t even understand genres!
And if you think that’s bad, get this: Udit Narayan is tagged as “Odia Bhajan,” meaning if you listen to him enough, your Discover Weekly is filled with Bhajans. To test this (for science), I actually did this, and it worked.
Spotify's solution to this was to leverage algorithms and data about listening patterns and creating smaller, very niche genres, and attributing these sub-genres to one artist. There’s over 6,000 genres, each weirder than the last. “Vapor Pop” is described as a combination of elements of vaporwave, synthpop, and chillwave. There’s also funk 150 bpm, neo mellow, talent show, shoegaze, bedroom pop, adult standards, and noise pop?! Social media also heavily influences what gets classified. This is how we end up with genres like “Viral Pop,” which exist solely because of data trends on platforms like TikTok. If some of these sound made up, that’s because they are.
While these niche genres attempt to capture some of the nuances of Western music, they often miss the mark with more diverse music from non-Western cultures, especially when applied at an artist level. Leaving everything to data and algorithms doesn’t always yield accurate data.
Glen Macdonald, the former lead data scientist at Spotify that made the niche genres, has also made the “Every Noise” website which uses machine learning to put genres on a spectrum purely using data (tempo, key, harmony, rhythm, and timbre) to cluster similar genres closer together. His approach to music classification relied on analyzing large sets of listening data to identify patterns, like how certain sounds or tempos resonated with listeners.
By using this data, Spotify was able to create new genres or "clusters" that reflected the evolving musical landscape. These clusters often overlapped, allowing for more dynamic and layered categorizations. Genres that were sonically similar were placed close to one another on the map, allowing users to explore how genres like synthwave and vaporwave are related or how genres like soul and R&B compare.
Now, because Spotify has its own genre situation going on, I had to manually map its data back to actual, universal parent genres. A shoutout to Chosic for helping me do that!