AI attempts to identify when music transcends genres

AI can now have conversations, create art, and even make music, but can it turn the subjective into the objective?

by WILL BETTS of Musiio Blog

When The Guardian recently claimed that London experimental jazz act Alfa Mist’s new album Variables transcends genres, we asked ourselves whether we could verify that assertion with data.

What would genre transcension entail? According to Collins, “Something that transcends normal limits or boundaries goes beyond them, because it is more significant than them.” Given this definition, we’d expect to find many genre influences on Variables, and for there to be some other glue that keeps the album together. 

To test exactly how genre-bending Alfa Mist is, we compared Variables with other artists hailed for their cross-genre appeal. Harry’s House featured nine genres, Sault’s latest album Nine contained 16, and Thundercat’s 23-track album Drunk featured 19. By this measure, Alfa Mist is undoubtedly taking inspiration from a wide variety of sources. That makes him incredibly hard to pin down. 

Another way of gauging Alfa Mist’s proclivity for bending genres is to see how many genres our AI ascribes his tracks on average across his album. In theory, the more genres the Musiio tagging AI gives a track.

Here are the results compared to those previously mentioned artists.

Artists with a higher average number of genre tags per track likely create more genre-bending music, rather than simply dipping into different styles from one track to the next.

Sure enough, with this analysis, it’s clear: Alfa Mist’s Variables truly lives up to its name, with a mean average of 3.6 genres per track. You could legitimately call that “genre-transcending”, albeit forgiving a little poetic licence.

With all this genre-bending, how is Variablescohesive?

With all that said, there are elements that tie Variables together. The first is that Jazz is still the dominant genre on the album. Another is recording quality. One astute YouTube comment on track three, ‘Aged Eyes’, sums up the unique sound of Variables: “Love how this somehow sounds like an obscure 70s record from Brazil but also really modern!”

Our Recording Quality classifier supports this observation. Seven out of the ten tracks on the album receive a Medium quality tag, consistent with other recordings from the 1970s. Other connecting traits include the use of only medium-energy tracks and minor keys.

Perhaps we are seeing an artist focussing on a vibe and using the genres as tools to deliver that vibe in different ways – like a chef adding spices.So, is Variables genre-transcending? It’s a resounding “yes”. Also, in an age where we’re constantly told our attention spans can be measured in seconds, Alfa Mist delivers a 45-minute through-listen that captures the attention track after track. And it’s largely the record’s genre-agnosticism that makes it so compelling.

To learn more about automated music tagging, and how descriptive AI can help your music business, drop a message to sales@musiio.com, or use the contact form.