There are a lot of people talking excitedly about big data and digital music, but relatively few companies who genuinely have a handle on the implications. The Echo Nest is one of them.
The Massachusetts-based company has been providing music data to a host of clients in recent years, including Spotify, Rdio, Vevo and Rhapsody, helping them to improve their music recommendations for users. Last week, Spotify bought The Echo Nest for a rumoured $100m: a deal that has made big waves in the music industry.
At SXSW today, The Echo Nest’s director of developer platform Paul Lamere – a well-respected figure in the music/tech world – presented his vision for “the future of listening”. It’s a vision that may well now influence the way Spotify evolves over the next few years.
“I prepared this talk long before I was an employee at Spotify, so this talk is not talking about any secret plans that Spotify might be having,” he stressed. “This is not a super-secret plan that Spotify is doing!”
Lamere suggested that 35 years on from the debut of Sony’s Walkman – a period that’s taken in the launch of CDs, Napster, iTunes and streaming music services as new stages in music listening – the way we actually interact with music hasn’t changed that much. We play, skip and repeat songs and build our own playlists.
“Technology can really affect how we as listeners interact with music, and we are in the middle of this incredible digital music revolution where any song is just a few taps away,” he said. “I think the next step in the music revolution is changing how we interact with music.”
Lamere added that not all music listeners are created equal: everybody has their own way of listening to music. Citing studies from 2003 and 2005, he divided listeners into four groups: savants, enthusiasts / engaged, casuals and indifferents.
Around 10% of music listeners are savants: their lives are wrapped up in music, and they spend $1,000 or more a year on it. 20% are enthusiasts: they love music, and perhaps spend $100 a year on it, but have other interests too.
30% are casuals, who quite like music but aren’t that engaged, and may spend just $10 a year on it. And the remaining 40% who fall into the indifferents category? “If music went away, they might not even notice,” he said. They might spend a dollar at most a year on music.
Lamere suggested that current digital music services have already captured the savants and enthusiasts. The next step for Spotify and rivals will be to go after the other 70% of the world, partly with pricing – “find a way to engage these casual and indifferents by giving them a free way to listen to music” – but also with new ways of listening.
“This is really a UX [user experience] problem now. Just like people are not going to want to spend the same amount of money, they want to have different levels of interaction with music,” he said.
Savants will put the time in to search for new music and create playlists. Engaged listeners may look for a few artists and Spotify, while casuals may hit shuffle-play in iTunes or listen to a personal radio service like Pandora. And indifferents may only experience music on the radio.
“You have to find a way of engaging these people who would be totally intimidated by a search box that sits in front of 30m songs,” he said, describing these casual and indifferent music listeners as “lean back” listeners: they don’t want to lean in and spend too much time figuring out what to play.
“The model for these music services is radio. The radio is the key place where these indifferents go, so the challenge is to give them a music experience that’s as simple as radio, but better than radio. If it’s not better, they’ll stick with radio,” he said.
One of the ways The Echo Nest has been working on this is understanding different listeners’ tastes and habits, including investigating whether it’s possible to make certain assumptions about the music they might like based on their gender, age or location.
Lamere has published three separate studies on these individual demographic data categories in recent months: gender, age and location. They each involved taking a large sample of data from digital music listeners and analysing patterns in how their preferences matched their demographic groups.
He admitted that there are dangers in generalising – for example, while Justin Bieber may be hugely popular among women, that’s a different thing to assuming that specific individual women will thus want to be played Justin Bieber by their digital music service of choice.
“Of course there’s a big red flag here: you can summarise a whole room full of poeple through statistcis, but when you’re describing an individual you’re gonna get into trouble. There’s going to be a lot of men who like teenpop and a lot of women who like gangsta rap,” said Lamere.
The point: all this demographic data can be used to improve the listening experience a little bit: if you know someone’s age, gender and location, there are certain artists and genres that may appeal more – although the point about individuals not conforming holds true still.
The real customisation, though, will come once people start using a digital music service: the data generated from the songs they play, repeat and skip provides a wealth of information that The Echo Nest uses to create “taste profiles” – which can then be used to govern what music its customers recommend.
Lamere said that the digital music services of the future will add this data into an awareness of what a listener is doing and where they are – context. “It’s Tuesday morning, give me music for waking up or getting confident for a meeting, or it’s Friday night, give me music for bedtime, or hosting a sexy pool party, or programming with friends,” he said.
Right now, this is hard to scale to individuals: one person may like heavy metal for getting up in the morning, while someone else may prefer Coldplay. Lamere suggested that the next step will be contextual, personalised playlists. “Not just a wake-up playlist, but a wake-up playlist that knows that I like classical music,” he said.
Devices like smartphones will also play a crucial role: their sensors enable them to know where we are, whether we’re running or walking, and other attributes. “You can imagine in the near future a music player where you hit the play button, and it knows ‘it’s Paul, it’s 10 in the morning, he’s in the coffee shop’,” said Lamere.
“Already the device manufacturers are starting to get excited about this... Our devices will know whether we’re excited, whether we’re stressed, and will play us the right music based on that.”
The latter part of Lamere’s talk focused on how to make music more engaging: giving people a way to interact with the music itself, rather than simply hitting a play button and listening to it. He showed a number of demonstrations created at music hack days by developers tapping into APIs from various music and technology services.
They included his own Cannes Burn visualisation for Ellie Goulding’s song Burn and the recently-released Rap Demonstration mobile app for Tinie Tempah, which enables fans to rap along to the star’s songs using his mouth.
He also showed Music Popcorn: a bubbly flowchart-style way to browse different genres. “The idea is to put the listener back in charge of the music discovery process,” he said. Meanwhile, Lamere said there is more work to be done on understanding how to lead music listeners from one artist to another, introducing them to new music along the way.
He used the example of plotting a Six Degrees of Kevin Bacon-style path from Miley Cyrus to Cannibal Corpse, showing an app called Boil the Frog, which takes around 15 steps to make that journey. Another, called Going Undercover, accomplishes it via cover versions, while a third – Roadtrip Mixtape – does it using physical locations.
Lamere showed some apps for playing with the music itself, too. One called The Bonhamizer was based on “the theory is that any song can be improved if John Bonham was the drummer” while another – The Infinite Jukebox – uses canny loop manipulation to play songs, in theory, forever.
Finally, he showed Girl Talk in a Box, which puts music into the browser, and lets the listener interact with it, moving forwards, backwards, stopping on beats, jumping around and generally taking control of the music.
Lamere demonstrated it with Skrillex at ear-splitting volume, making the point that while these hacks may seem silly, they are early efforts at interactivity that could turn casual listeners into enthusiasts in the future.
“The idea is to make apps that really engage the casual listener and make them really want to dive into a song,” he said. “We have 30m songs in our pocket, we have to do a lot more than putting a search box on top and saying ‘go at it’. We have to engage these listeners,” he said.
“Everyone likes music, I think, given the chance, given the education, and given the tools to dive in and have fun with music.”
guardian.co.uk © Guardian News and Media Limited 2010