Meet the website that maps every genre in the world

Brought to you by Heineken

Heineken. Open your city.

We’ve teamed up with Heineken to introduce you to some men and women who have chosen to go beyond their borders, challenge the status quo, say ‘why not?’ instead of ‘it can’t be done’ – and as a result have made the world a more interesting place for the rest of us. For more people worth watching, head here.

In conspiracy movies there’s always that scene where someone makes a ‘paranoia wall’ out of newspaper clippings and whiteboard scribbles, everything connected to everything else in an arcane pattern that seems like madness. The musical equivalent is Every Noise At Once, a website that collects genres in a visual tangle, stretching from hardcore techno to vintage gospel via Japanese jazztronica.

Related styles like drum’n’bass, house and jungle all cluster together, but pick a direction and you soon find yourself at speedcore, moombahton, or sunset lounge. Click on a genre and you can hear a sample linked from Spotify, and then a sub-map of artists within that genre. It’s possible to clamber down each rabbit hole as far as you want to go.

Every Noise At Once a fascinating attempt to map something that’s fundamentally unmappable. No two fans of any given musical subgenre will agree with what best represents it or which other genre it’s closest to or sometimes even what its real name is. But the map’s creators – a group called The Echo Nest – only interfered in the form of tweaks and corrections and decisions between “nu soul” and “neo soul”, with the rest of Every Noise At Once created by datamining user descriptions on Spotify.

Originally an independent group, The Echo Nest were acquired by Spotify, which has given them direct access to the user data they need for visualisations like their follow-up, Genres In Their Own Words. Rather than a map, it’s a pairing of each musical genre with a list of the words that appear with most uncommon frequency in the titles of songs labelled on Spotify as belonging to that genre. Some of the lists read like a kind of jumbled poetry. Here’s disco, for instance:

That’s disco all right. All the restrictions of “don’t” and “can’t” exploding into yearning “let’s” and “want” and “wanna give feel”. Meanwhile, this is pop-punk:

I’m pretty sure “summer hate dead” is an actual emo song and “this your american kids song” could sum up almost all of them.

While it’s fun to bounce around in these visualisations, they also serve a useful purpose. Every Noise At Once is a great way to find new music you might like by picking a spot on the map that aligns with your taste and exploring from there. Even choosing at random can lead to new discoveries, new things to add to your playlist. Maps broaden our horizons, and by telling us where things are they immediately make us want to visit them.

Glenn McDonald is The Echo Nest’s resident data alchemist and it’s his job to cook up these ‘paranoia walls’: these maps of music’s furthest territories.

How long have datamining music for?

I’ve been using software to organise music information as long as I’ve been using software, which would be since about 1985, but I think the first thing I did that could be called “datamining” was an analysis of the voting patterns in the Village Voice’s Pazz & Jop music poll in 2001.

What inspired you to follow this as a career?

I didn’t exactly “follow” it. I spent a lot of years designing information software for various purposes not directly related to music, but I would always end up using the software for my own musical interests as well. The transition from ‘music as a way to demonstrate software’ to ‘software as a way to understand things about music’ was a fortuitous and belated bit of serendipity, after one of my previous employers got acquired and that particular data-analysis project got shut down.

How much work is it to put together something like Every Noise At Once?

It’s a large amount of my work that builds on a vastly larger amount of work done by several dozen other people over the course of many years, starting with Brian Whitman and Tristan Jehan forming the Echo Nest. And even beyond that, the data we collect is essentially the work of hundreds of millions of people listening to music and writing about it and sharing it and caring.

Why do you think it is that dance music has so many subgenres? Do you have a favourite?

Dance and metal are particularly rich veins of subtle distinctions. I think in both cases the relatively strong borders between non-metal and metal, and between non-dance and dance, allow the internal distinctions to flourish in protected conditions. I like a lot of EDM forms, but the one I orbit back to most frequently and instinctively is probably Progressive Uplifting Trance. Preferably something with Emma Hewitt singing.

Did you discover new music you were into while working on it?

Always! Sometimes many whole astonishing genres in a single day. There’s so much more amazing music than most people have any idea. This is what drives me to keep adding and refining. Your favourite song in the world is probably a song you have no idea even exists. I’m trying to find all the songs everybody has no idea they would love, and organise them so they have a prayer of being found.

The disco section of Genres In Their Own Words is a favourite of mine, it sounds like a fractured overheard conversation. Do you have favourites? Were you surprised at how some of them turned out to be a kind of poetry?

That whole thing was an offhand experiment I tried with no idea if it was going to be interesting at all, and it succeeded well beyond any of my most optimistic expectations. You can just start reading anywhere and it’s uncanny and hilarious and wonderful and telling. I did a very small amount of technical work to make it happen; the wonderfulness is pretty much all the work of the world.

Which of your findings has most surprised you?

Before starting the map I knew that there were hip hop groups in other countries, but I had not even the vaguest idea how many regional scenes there were or how deep. Australian Hip Hop was the one that first made me suspect how much I didn’t know, but pretty much everywhere you look, you find people picking up their own microphones and doing something that’s not quite the same as what inspired them.

That’s true in pop and rock and metal and many other things, as well, but I think hip hop is so uniquely focused on language that a) local cultures demand their own hip hop even where maybe they can get by on imports in other styles, and b) hip hop in “foreign” languages can seem much more foreign than the “foreign” versions of other styles.

So as a result, I think, even most hip hop fans are exposed to very little “foreign” hip hop, and thus inevitably have little idea how much hip hop there is everywhere else. Witness how hilarious US hip hop fans tend to think Grime is defined by the rappers having British accents. And that’s not even technically a different language.

Are you aware this isn’t a normal thing to do? Are you in fact slightly crazy?

Oh, I’m hardly the first person to try to understand or categorise all the music in the world. I just came along at almost the first point where it was possible to do so at scale, and with music samples. Which doesn’t prove I’m not crazy, of course. But I don’t feel crazy. More like stubborn, patient, and tortured by the idea that there is even more music I haven’t heard yet.

What’s changed about your work since the Spotify deal?

We now have more and better data, on one side, and are much more intimately connected with the products of that data on the other side. As an API company, both our inputs and outputs were more indirect and filtered. For one particularly useful example, at Spotify I know where every Spotify track is being played, geographically. At The Echo Nest we often only knew that a track was being played, not where. So I could not have done a thing like Every Place at Once before now.

What are you working on at the moment?

My latest new thing is The Needle, which is another in a long series of attempts to use listening data to detect hot new emerging tracks as they are still emerging, and another example of where geographical data lets me do things I couldn’t do before. The Needle’s predecessor was The Echo Nest Discovery, but the END was just one global list.

One global list is kind of fun, and there’s a global version of The Needle, too. But then there are also individual versions of the Needle for every country. What’s new and hot and emerging in Brazil, but unknown elsewhere? Or in Estonia, or Taiwan? Before now, I couldn’t effectively find out. Now I can. And, of course, it turns out that everywhere you listen, you find more wonderful weird music you wouldn’t otherwise ever hear.