Arcona uses machine learning to compose adaptive soundtracks in real time

Published December 11, 2019

Arcona Music took to the stage at Disrupt Berlin today to showcase its adaptive music service. The local startup utilizes machine learning to create musical beds capable of adapting to different contexts in real-time. The user simply needs to input a handful of parameters, and the service will adjust accordingly.

“Give it a style, an emotion and a musical theme, and you can say, ‘play this,’ and the engine will take that blueprint and realize it,” service cofounder Ryan Groves explained, in a conversation with TechCrunch. “If, at any point, the emotion or style changes, it will adapt to that and create this essentially infinite stream of music. You can play a particular song blueprint for as long as is necessary in any dynamic environment.”

The service is still in its infancy, at the moment. Its two founders are its only two full-time employees, along with a part-time developer. Groves and co-founder Amélie Anglade bootstrapped the scrappy startup, which has yet to seek funding.

Groves is a composer and musical theorist who formerly worked at popular AI-based music composition service, Ditty. Anglade is a music information retrieval specialist who worked at SoundCloud.

Rhythm gaming is the first clear application for the service. The popular gaming genre is built around a changing soundtrack and could potentially benefit from music that requires minimal pre-programing. Moving forward, the potential for such a service is far broader.

“In the very long term,” Groves said, “we should see this being almost your own personal orchestra, leveraging augmented reality, GPS and all that stuff, and just responding to your environment as you’re listening.