Are Algorithms In Tune With Music?

What impact do algorithms have for music curation and creation?

There’s so much talk these days about Algorithms and Artificial Intelligence and scientific innovations have not only become a part of our everyday lives but our everyday language. But if you walk down the street and ask a random person ‘What Is An Algorithm?’ it’s unlikely they’ll be able to give you a straight answer. So when we got the opportunity to partner up with The British Science Association to explore the role of algorithms in music, we couldn’t wait to dig into this one.

Algorithms are found everywhere nowadays, in fact they’ve existed for thousands of years, which to be honest wasn’t something we even thought was true, until Alex McLean co-founder of Algorave explained how, “Thousands of years we’ve made algorithms, you don’t need a computer to make an algorithm. The whole basis of a woven material, involves algorithms of interacting systems, it’s all about patterns really. I guess at this point, things are changing because algorithms are coming into our social lives in ways they haven’t before. But the basis of an algorithm is just mathematics and we’ve always been doing mathematics.”

But beyond maths, algorithms are being used more and more in music curation and creation, from generating playlists of our most listened to tracks to even creating background music for games and apps – how much do we really know about their impact on the music industry and musicians?

These days if feels like Algorithms are changing our lives at breakneck speed. We’ve moved past the disruption that digital formats first wrought on the music industry, and its back to business as usual. Infact it looks like the music business is booming and streaming is giving us a lot of reasons to celebrate. So why all this simultaneous excitement and nervousness about Artificial Intelligence and Algorithms – should we be worried that one day algorithmically led curators and creators could replace DJ’s and musicians? Are algorithms leading us or are we leading them?

 

The BSA’s vision is of a world where science is at the heart of culture and society and British Science Week (10-19 March) is a perfect time to showcase some of the unusual places science can be found. Nation of Billions in collaboration with the British Science Association have produced a short film exploring the concept of algorithms and their use in music.

There’s a lot to unpack when it comes to algorithms in music and we take a look to see if they’re really in tune with music creation and curation? DJ Semtex explores the role of algorithms, talking to data scientists and industry executives working in music. The short film features interviews with BBC Radio 1 & IXtra’s MistaJam & 1Xtra’s Jamz Supernova; Spotify’s Austin Daboh; Jukedeck Founder – Ed Newton-Rex; and co-founder of Algorave – Alex McLean.

So could an artist actually ‘break’ Spotify? Well apparently Ed Sheeran almost did. After securing the biggest first week streams ever, Spotify released a statement saying “It’s fair to say that Ed Sheeran nearly broke Spotify this week.”

Ed Sheeran famously once said that he ‘owes his career’ to Spotify. So of course we had to get in touch with Austin Daboh, Senior Content Editor at Spotify and ask him about the impact of Spotify’s algorithmic curation through playlists and how they can benefit an artists career. With back to back record breaking releases from Stormzy and Ed Sheeran, streaming numbers have been big news recently, Austin told us what that really means in real terms, “in terms of UK artists having international appeal, right now as we record this interview, Stormzy’s ‘Cigarettes and Cush’ is sitting in Rap Caviar in the 10th position, it’s a really high position, and opportunities are being given to UK artists on a global level that didn’t exist before Spotify was around.” 

Playlists are the fundamental business of Spotify and they seem to have perfected the playlist game, but its not an either algorithms or humans, type of decision? It’s a combination at Spotify, as Austin gives us some more insight, “We have 3 different types of playlists on Spotify, we have 100% handcrafted curated playlists, we have what’s called ‘Algotorial playlists’ which are Algorithm based suggestions that are then hand-picked and then filtered down into a smaller batch of records by human curators. So its half-picked, but then the wider batch of records in which those records are hand-picked from is spurted out by the computer and that’s based on listener tastes and habits and then we’ve got 100% fully algorithm based playlists.”

So we have to try and find really strong and smart ways in which we can offer a personalised service to our users globally across the world.

“We have over a hundred millions users worldwide, so for us to create a personalised playlist for each one of those users it’s not possible. So we have to try and find really strong and smart ways in which we can offer a personalised service to our users globally across the world. In terms of how that’s decided, it’s what people are listening to, the genres they’re listening to, the types of artists that they’re playing, how often they’re playing them. There’s a big bunch of different data points that are looked at when the algorithm’s working out what someone might like.”

Spotify launched their first personalised algorithmic playlist in July 2015 and it’s impact has been significant, Austin breaks down how they work, “I think that there’s no doubt that our algorithm based playlists have been really successful, if you look at the usage of Discover Weekly which is the first one that we launched and then Release Radar – Discover Weekly gives you a mix of new records plus some old records it thinks you might like. Release Radar is 100% records that have been released that week. In terms of the usage, consumption of both of those playlists its been proven to be quite successful.”

The growing size and scale of streaming services have undoubtedly been driven not just by their users, but a powerful machine-learning system and human content editors like Austin, “With Spotify effectively we’ve got a birds eye view on that same level of data just on a much wider scale. I personally don’t think you’re ever gonna beat hand-crafted human curation but when you’re dealing with the numbers of subscribers and users that we’ve got, hundred million plus users worldwide, then you definitely do need data driven help, to help mix in that hand-crafted curation with what the computers showing you that the data’s pointing towards.”

With Spotify’s growing user base, data and personalised playlists, both Apple Music and Tidal are also expanding human curated and algorithmic playlists. Apple Music recently launched it’s ‘My New Music Mix’ while Tidal enlisted music journo Elliot Wilson as the Editorial Director of Culture & Content where he’s already kicked off by launching a number of curated Hip Hop playlists.

Artificial Intelligence (AI) being used in music making is an area that’s started to generate a lot of intrigue and UK startup Jukedeck are building an artificial intelligence system that “understands the rules of music theory and can use that understanding to write its own music to help contribute to other peoples music and to generally get involved in the music creative process.”

It’s also arguably one of the more controversial areas of the industry for many musicians who feel that AI’s can never replicate the emotional input from human creation. Ed Newton-Rex the founder of JukeDeck however doesn’t discount the human factor and looks at AI more as another tool for music makers.

I don’t think we’re going to get to the stage where we only use or listen to music that is made with algorithms – Ed Newton-Rex, Jukedeck

“I think that it’s great that we have this whole spectrum of possibilities. But even within music that involves algorithms I don’t think that there’s too much of a danger that all music will end up sounding the same” Ed compares AI’s to the way that software like Ableton Live and Logic have enabled a new generation of bedroom producers who are literally able to make music in their bedroom. Seen more as an aid in the music making process, Ed explains that “actually algorithms can help people to experiment with new sounds and new ways of doing things. They can form the basis of computer programs that can lead in fact to more variation in sound. So, I think algorithms will expand the variation that we have across music.”

Our different emotions are channeled into the way we listen to and the way we make music as humans, so can emotion really play a part with AI? Ed, illustrates how the combination of humans and algorithms can work in reality, “I think that’s where it gets really emotional is where people add to the music. So probably the most emotional I’ve been seeing this music being used is when I saw a video of a child who’d used our music as a backing track and who’d sung this beautiful song she’d written over the top. So she didn’t have the ability to write that backing track herself but she sung this beautiful song and suddenly you have this combination of a person and a machine, sort of enabling this creativity and I think then, the emotion really shines through. AI hasn’t got to the stage where it can replicate human emotions and that kind of thing but I think when you bring people into the mix and get them to use algorithms that’s when you get the emotions and all the brilliant aspects of music that we really value.”

But while the debate about the abilities of AI’s are only starting, will they ever have the ability to rancour with our emotions the way that Ed Sheeran does? Like his album ÷ or not, the question about whether lyrics written to popular keywords could actually be replaced by an algorithm itself, and it actually got us wondering – could an algorithm really write a song like Sheeran?

To kick-off with our first Live Event Series of 2017, Nation of Billions are partnering with the British Science Association (BSA) to host a Q&A. This time we’re ‘Droppin Science’ and asking “Are Algorithms In Tune With Music?”

This is an exclusive and transformative evening event, which will give you an opportunity to hear leading music industry executives and experts discuss about their experience and views about the future of an exciting new era in music curation and creation. Featured guest panelists on the evening will be DJ Semtex, Jamz Supernova and Ed Newton-Rex, and the event takes place on Tuesday 28th March at Huckletree Shoredtich from 7-9pm.

Earlybird tickets are on sale now for a limited time only, get them here now. 

Find out more about British Science Week: www.britishscienceweek.org

Find out more about the British Science Association: www.britishscienceassociation.org