Most of the meat and bones of this post is in the visualisation below so I’ll skip the usual lengthy preamble and just provide some more context and personal observations below the viz. Enjoy!
I’d always been interested in the Spotify API, particularly the Audio Features and Analysis it provides for every track. I think it’s pretty likely that Spotify uses these features to create personalised playlists such as Discover Weekly. There are so many things you could do with this data but as a start I wanted to see if it would be able to reflect geographical differences in music taste. I used Spotipy, the Spotify API wrapper, to extract the audio feature data for each track within a country’s top 50 playlist such as this one for the UK. Note that these playlists are actually charts and are constantly updated.
Looking at Danceability, I found that it didn’t vary a whole lot among countries. Most of the Americas and Europe averaged around 0.7 on the dance scale except a few noticeable exceptions such as Brazil. It could be that their Top 50 playlist at the time didn’t reflect their love for dancing or maybe they just don’t listen to a lot of danceable songs on Spotify. Brazilians, feel free to reach out to me and clarify!
Energy and Valence is where South America really shines as they’re the reddest and yellowest continents on the map. This didn’t come as too much of a a surprise to me as Latin music is typically high energy and joyful. Additionally, Spain was among the top 5 when looking at both these features so it may be a preference among Spanish speakers and not just geographical location. The USA was pretty low in terms of Energy and Valence probably because at the time of pulling the data, the playlist was dominated by Drake and… you know how that usually goes.
I would interpret the Speechiness data with a grain of salt. It ranges from between 0.05 and 0.17 which doesn’t match Spotify’s description where it says,
“Values between 0.33 and 0.66 describe tracks that may contain both music and speech, either in sections or layered, including such cases as rap music.”
The USA’s Top 50 chart consisted mainly of rap and hip-hop tracks but it still averaged 0.1588. While the numbers may not be very reliable, I think the differences between regions are still telling. You thought Drake was loved in the USA? The top 10 tracks on Bulgaria’s playlist were all Drake; it had the highest average Speechiness.
Finally, please note that this is not even close to a complete understanding of geographical music tastes. The data only accounts for what Spotify has deemed popular among its users in certain countries at a certain point in time… which makes me think I should have chosen a less Drake time to pull the data. Anyway, please don’t take these findings super seriously! 🙂