Two separate announcements yesterday highlight the fact that big-data and machine-learning for A&R purposes are enjoying a bit of a moment – after years of being mostly seen as a controversial niche when they cropped up in the on-stage conversations at industry conferences.
Music Ally has been to countless panel sessions where A&R veterans scoffed at the idea that any kind of algorithm could ever rival the “gut feeling” of experienced human talent-spotters.
The counter-argument has always been that this isn’t the goal: but that these kind of systems can complement those skills, and perhaps also provide solid support for those talent-spotting guts.
That certainly seems to be the instinct behind Warner Music Group’s decision to acquire a startup called Sodatone, whose technology draws on streaming, social and touring data to spot unsigned talent with early momentum.
We first wrote about the US company in May 2016 when it launched in private beta, with its promise to let labels, bookers, promoters and managers “know if an artist is spoofing data and how passionate their fans are”.
Its founders will now report in to WMG’s chief data officer Vinnie Freda, and work with teams across WMG and Warner/Chappell. The hope, from WMG’s recorded-music boss Max Lousada: “Sodatone will help to differentiate us in the search for the superstars of tomorrow.”
The news came as another WMG-linked startup, Instrumental (founded by former Warner exec Conrad Withey), announced a $4.2m funding round to continue developing its TalentAI technology.
It scans “influential playlists” as well as social platforms with the same aim: to spot emerging artists showing early signs of (genuine, not fake) momentum, for labels, promoters and publishers. In his pitch at the Midemlab startup contest last June, Withey said Instrumental was generating £100k-£150k a month from its business.
Warner Music was one of the investors in Instrumental, so its decision to acquire a rival could be seen as a disappointment for the startup. That said, the Sodatone acquisition could also fire the latest starting gun for consolidation in and around big-data / machine-learning firms with proven technology – with A&R the focus this time rather than (as in the past) analytics for artists.
Meanwhile, what happens with Sodatone now will be important to watch: how can this kind of technology develop within a label and publisher if used on a day-to-day basis by A&R teams?
Just as the most interesting thing about artificial-intelligence music-composition technology is how human musicians use it, so the most interesting thing about data-based A&R is how those human talent-spotters (and their guts) use it.