"What music do you like?" One way to sufficiently answer this notorious smalltalk question

I am sure, you have been asked "Hey, what do you listen to?" a million times. Some people can easily answer that question. I never could. Until now. Thanks to the new digital data collecting behemoths that record our every moves, I can tell you exactly, what music I listen to! I am a Spotify user and they told me at the end of 2019:

A depiction of the genres, I listen to most, as
                        compiled by Spotify:
                        Electro House, Dance Pop, Rock, Hip hop, Big beat.

When I saw this, I was thrilled. Finally an answer, depicted in a nice diagram. A hard question, broken down to just five genres, even showing me, how much I like each of them. Just, that this picture does not. My thrill puffed away, as soon as I saw a few of my friends' diagrams. They looked exactly the same. Sure, the genres were different, but the graphic shows the same distribution, the same five bars with the same heights. Spotify cut some corners and did not use their vast amount of data to satisfy my desire for insight.

That did not sit right with me, so I decided to take upon that issue myself and I came up with my own graphic:

A depiction of the genres, I listen to most, as
                        compiled by me:
                        Electronic, Big beat, Pop, Downtempo, Dance pop.

When looking at this graph, it becomes clear, why Spotify went with their designers instead of their data analysts on displaying this graph. But still.

While this fun little graph, now upholds to my pedantic standards, it does not really answer all the questions I have. That's why I went totally overboard and created a four-dimensional space of all the genres I listen to (at the bottom of this page). But where did I get the data from, and how can you create your own multi-dimensional genre map?

Tldr; Take me to the visualization

When the European Union slowly realized that it is loosing the race in internet behemoism, they decided, instead they could try to stop the user data export to the United States. This was attempted by introducing a new legislation, the well known GDPR ("General Data Protection Regulation"). It was introduced on May the 25th, 2018 and is one of multiple legislations that aims at protecting consumers online. And the best part: With virtually no international web-based giants, the introduction of GDPR would not even hurt the big European web companies (because there are none). On thing, GDPR does: It makes it mandatory for companies to provide their users with the ability to receive all the personal user generated data on request. In the case of Spotify you can just download all the songs you liked as a handy machine-readable file. That is what I did. And I analyzed it.

For plotting my genres onto a multi-dimensional space, Spotify plays a second role, as well: In 2014 Spotify bought "The Echo Nest", a company, focused on music analysis. Glenn McDonald, an employee and data engineer of "The Echo Nest", made an affort to analyze into what genres the music on Spotify can be devided and how these genres relate to each other. If you haven't seen it yet, you should totaly check out Glenn's site everynoise.com. Over there, all the genres that Glenn's analysis produced are plotted in relation to each other, with bands and songs of these genres, available with just a few clicks. Even if you are not as much a data-visualization-fetishist as I am, it is worth a look. Not only does Glenn track over 3.000 genres, he also provides many playlists for all these genres and a lot more. Go. Check. It. Out. Then come back here... ;) If you have, you will notice a slight resemblance with what you see on the bottom of this website, but more on that in a sec.

What has Glenn's effort of tracking genres to do with us? After scraping the genres, I can put them into relation with the music of my own liking by just show the genres, I listen music from. And these are displayed in different sizes, based on how many songs in that particular genre I have liked on Spotify.

Surprizing Insights: I listen to "lgbtq+ hip hop" Wait, thats a genre?

Some insights are in sight. The data is presented in the same way as it is on Every Noise. About the orientation from left to right and top to bottom I will let Every Noise speak for them selfs:

"The calibration is fuzzy, but in general down is more organic, up is more mechanical and electric; left is denser and more atmospheric, right is spikier and bouncier".

Turns out, I like quite a broad spectrum of music but I do have a focus on the electronic, hip-hop, rapish, poppy, rocky music. Below are my top 15 genres with the amount of songs:

GenreAmount of SongsPercentage of my total songs
Electronic 704 22.4%
Big beat 493 15.7%
Pop 317 10.1%
Downtempo 311 9.9%
Dance pop 308 9.8%
Trip hop 303 9.6%
New rave 292 9.3%
Rock 292 9.3%
Electro house 288 9.2%
Edm 264 8.4%
Hip hop 231 7.4%
Alternative dance 216 6.9%
Rap 212 6.8%
Breakbeat 197 6.3%
Alternative metal 196 6.2%

My music collection does offer some weird outliers. I was amused to see that I apparently am listening to "lgbtq+ hip hop". Apparently, Azealia Banks is responsible for this lable. Yes, that rough racism tweeting music marvel who raps about loving white dick (click to play: Liquorice) or here iciness (Ice princess).

It also becomes apparent, how international music can be nowadays. I listen to Slovenian electronic, Brazilian electronica, Swedish (synth | synthpop | electropop | pop | indie pop), French hip hop, Australian, Icelandic, Mexican, South African, and Austrian pop, South African alternative, Cape Town indie, Russian (rock | trap | hip hop | pop), British (insert long list of genres here), German (metal | pop | punk | rock | hip hop | techno | baroque), and many US American genres. The list goes on.

Now, let's take a look at the genre space! (Depending on your screen size, some scrolling might be required.)

When you hover over a genre you can see the artists that I like within that genre. Clicking on them (most of the time) gives you a preview of one of my liked songs of them. It also then displays a link with a little ♫ that brings you to the song on Spotify.

The data is not perfect. For example, I am not a big fan of christmas music but because Frank Sinatra, Elvis Presley, The Kingston Trio, Tom Jones, and Gene Autry have been invested in producing some christmas songs this genre pops up for me. Pop is overrepresented in my data due to the fact that many of the big artists produce pop music besides music of more specific genres. Even when I listen to the non-pop songs only, the algorithm behind this graphic picks up pop, as it is associated with the artist. This does not prevent the bigger picture to emerge: You can see clusters around house/techno/electro-ish music, trap/big beat, dance, hip hop, metal/rock and pop.

And now: It's your turn

Should you, surprisingly, be more interested in your own taste in music than in mine, you can go ahead and do the same: You can load your own Spotify-data and scroll through it. In order to download your Spotify-data you need to go to your account page, scroll down to the bottom and request your data. It takes some time until Spotify provides you with your data (a few days) but then you get a handy zip archive with a file in it, called "YourLibrary.json". That's where the gold is. Btw, I think it is a good idea to download that data to prevent of you loosing your memory of what music you like, in case Spotify messes up your data by accident or shuts down or... you know.

Once you got your data, click the "Process your file" button, which does not upload your data anywhere. Pinkie promise!