Sony Music, The Orchard and Amazon Web Services held a machine-learning focused hackathon last week. Music Ally couldn’t make the event, so we asked if Sony could give us some information about the winning hacks. The label group came back with a full report on the event, quotes’n’all. So, in its own words, here’s what went down at Music ML 2020.
“Music ML 2020 is the second annual hackathon held by Sony Music, The Orchard and Amazon Web Services (AWS) and the first to be expanded out of the US. The Orchard office hosted the New York competition and Sony Music simultaneously hosted in London over the course of three days. The hackathon’s primary purpose is to allow cross-functional teams of industry professionals to create business solutions at the intersection of music and machine-learning.
Each team was tasked with creating a proof-of-concept for a new machine-learning solution that would positively impact the business. At The Orchard, teams worked daily to connect fans to songs, albums, and videos from their favourite artists across the globe.
The hackathon focused on improving the quality of content releases and making them more discoverable, using machine learning as a toolset to support creative decisions. The opening keynote by Jacob Fowler, Chief Technology Officer at The Orchard, challenged competitors to inspire, create, and push the boundaries of innovation.
In each location the teams were comprised of technical staff from Sony Music, The Orchard UK, and solutions architects from AWS. BRIT-nominated producer DJ Fresh, who also works as a machine-learning engineer, joined the British hackathon to lend his expertise.
On the competition’s final day, the UK teams presented their projects followed by the New York teams.
Five executives made up the UK judging panel. From Sony Music UK: Cassandra Gracey, President, 4th Floor Creative Group, Michael Hanson, Head of Digital, Columbia Records and Olivier Parfait, Director, Global Business Development & Digital Strategy. From The Orchard UK: Chris Manning, General Manager, UK & EU and Joe Andrews, Senior Director, International Sales and Marketing.
The US judges were Jacob Fowler, CTO, The Orchard; Rachel Stoewer, VP, Artist and Label Services, The Orchard; Devki Patel, Strategy and Finance; and Chris Frankenberg, VP of Emerging Technology, Global Digital Business, Sony Music.
In London the winning team design-trained a machine learning model to understand and monitor user behaviour and interactions with brands and artists. This concept put a powerful and user-friendly tool in the hands of marketers, empowering them to rapidly create retargeting lists comprised of high-value users.
This is particularly valuable for breaking new artists, as the model can identify high-value users with only a small amount of input data. The tool can also be used by sync and brand partnership teams, enabling them to actively pitch brands on potential deals.
In New York, the winning team produced a model that uses machine learning to auto-generate artwork for artist and label merchandising. This innovative design created interesting and creative opportunities for artists and labels to utilise existing content on various platforms.
Lewis Donovan, lead web developer & Music ML project lead for Sony Music UK, said:
“After three intense days in London the teams produced a tool that analyses data to inform A&R decisions, a spike analysis tool and an artist/brand/user affinity tool. Achieving this is just three days, by teams with varying degrees of coding experience illustrates the energy, creativity and ambition for innovation here. Every participant in the UK event had the opportunity to collaborate with new people, learn new skills, and help create a new business application using machine-learning.”
DJ Fresh, Ministry of Sound artist and machine learning engineer, said:
“The way we consume media now is three dimensional, as an artist I want to learn how to use that palette. The last few years, being an all-or-nothing kind of guy, I devoted myself to AI and trained as a machine-learning engineer, so when Sony Music asked me to get involved I thought it sounded like an interesting challenge.
I’ve always been involved with tech – in the early 2000s I started a site that was the biggest music forum on the net at the time, a place for the drum and bass scene to network.
Now I’m working on a new app called Golddust (Golddust.io), and with my new single Drive out now, it’s a very creative time for me.”
Cassandra Gracey, President, 4th Floor Creative and Music ML judge said:
“I loved seeing what could be achieved in such a short space of time, looking at ways to boost our business using various machine learning applications and our owned data. I look forward to the winning team’s programme being finished so I can start using it!”
Sony also noted that the winning UK team included Sony Music UK’s web developer Lewis Donovan, Francesca Lamaina and Josh Rubner, AWS’ Andrew Morrow, and (via a Hamburg dial-in) The Orchard’s Dino Celotti. The winning New York team was The Orchard’s Anthony Khoudary, Jinny Yang, and Peter Iannone, and AWS team members Julia Soscia, Rahul Popat, and Alex Jestin Taylor.