Classified: How Muru’s A.I. Turns Metadata into Perfect Track Sequences thanks to Proprietary Music Classification Solutions

There’s an opportunity buried in the world’s mountains of music metadata: classification, the principles that define genre and that rely on cultural context.

Classification is the big piece missing from the puzzle of how to deal programmatically but accurately with the deluge of digital music tracks, when the listening habits and sound cultures of the world vary widely. Classification powers curation and discovery, and thus income for music makers and rights holders.

Muru Music has trained machines to address classification, and this A.I. lies behind its remarkable ability to generate fluid, listenable playlists on top of music streaming services. Going beyond recommendation, Muru’s technology has translated a seasoned DJ’s brain into algorithms and rule sets. It digs deep into the planet’s millions of tracks to find new, cool, sonically appropriate music.

Muru came from founder and CEO Nicc Johnson’s experience as a professional DJ and resident, for one of the world’s groundbreaking clubs (Pacha Ibiza). He also worked as a consultant and crafted 24-hour playlists for restaurant and retail clients. He found himself profoundly frustrated by the tools available to him, as he plumbed his knowledge for unique, flowing track lists.

“Every platform had their own type of song classification and it was shoddy at best,” recalls Nicc Johnson. “It was especially evident when looking at more niche music genres like Deep House and Nu Disco or the subgenres of Jazz and Classical. After looking into various recommendation services, I realized that needed to be fixed first. Although a music playlist and discover app was the initial plan, it quickly turned into a music technology solution.”

“With Muru we have developed the first A.I. powered classification engine that can classify any song of any catalog automatically into the correct genre, more accurately than a human and a 1000 times faster,” using existing metadata and basic sonic parameters like bpm combined with a healthy dose of music theory, the fundamentals of Djing and the use of musicology. Johnson continues: “Our classification API could save companies like Spotify and Pandora millions a year, while generating more income for artists and labels, as Muru digs up a much greater number of less well-known tracks for listeners. It combs deep catalog and finds the gems there”

This classification allows streaming platforms to recommend music from the entire catalogue of songs, not just a small percentage of what’s actually available today. It turns a disparate, huge catalog into playlists that actually engage and flow. Playlists are key to the streaming music ecosystem and are the biggest form of music consumption today. A recent survey by musicwatchinc.com confirmed that 94% of Spotify and Apple Music listeners prefer creating their own playlists. Making your own playlist is incredibly hard, time-consuming and utterly frustrating. DJs learn to do it well and relatively quickly, using a complex but distillable rule set. Muru has captured these rules in its algorithms, making playlist creation close to effortless, while keeping it highly rewarding.

“A DJ specializes in two things: song selection and timing. Muru has music algorithms that think like a DJ and that let you create a playlist in seconds,” Johnson notes. “The combination of our classification and music algorithms means we can find contextual relationships between songs across all genres, allowing for a much better discovery experience.”

The Muru engine can keep pace with digital music’s unflagging releases. And Johnson and his team are making Muru even stronger, increasing accuracy for new and challenging genres that are highly underserved across all music platforms (jazz, classical, metal, electronic music) and building tools for machines to classify any genre from any musical culture around the planet, but also predict emerging genres before the market can spot them.

“We think our technology could change the way streaming services recommend music and the way people listen to the world’s recordings,” muses Johnson. “We’ve created something that can make sense of all the data and make something beautiful, by returning to the complex rules that allow for classification.”