We recently had the pleasure of discovering Midtempo Radio – an experiment in harmonic audio broadcasting that plays great tracks from a diverse music library as smoothly as possible.
Each track is analyzed on key, tempo, and intensity, then passes through various playlist selection rules – nothing about genre or era, and no A-list/B-list hierarchies.
You don’t know what the system will select, but often the music flows so smoothly and naturally you won’t notice when one song ends and the next begins.
The global playlist includes Electronic, Funk, Latin, Afrobeat, Soul, Downtempo, Modern Classical, Pop, Folk, Disco, Dubstep, Alternative, Reggae, Dub, Doo-wop, Jazz, Hip-hop, Blues, Balkan Beat, Indian Devotional, Bluegrass, and more.
Best part: There are no ads, sponsorship, or interruptions.
We caught up with Chris Burford, the man behind Midtempo Radio, to discuss his project, music streaming, and much more.
What was the jumping off point to start Midtempo Radio?
The upside of being in semi-isolation over the summer was that I gave me plenty of time to develop an idea that I have kicked around for years; a harmonically-driven station playing songs without genre or era constraints. Basically that means that the system selects the next track based on what it believes will sound best, rather than some arbitrary rotation rule.
I spent decades working both in radio and online, have DJ’ed at big festivals, have a foolishly large and eclectic music library, built big, complex database-driven systems, and finally found some free time… so creating an AI-driven harmonic station just had to be done.
What are the features of your product that distinguishes it from other streaming services, online radio stations?
There are probably other streaming services that use harmonics as a core part of their programming, but I’ve yet to find any that shout about it. Harmonic programming doesn’t really feature at all in the radio industry, yet it’s a common part of club DJing and has been for decades.
And yes, the quality of the playlist is important – no one will listen if the selection is rubbish – but it’s the space between the tracks that I’m finding fascinating.
Each song is analyzed on key, tempo, intensity, intro/outros, and depth, then, using that data, the system dynamically selects the next track based on a complex set of rules that should mean that the music flows naturally.
The library is willfully eclectic – from 1940s Swing to 2020s Electronica – as it’s the graceful combination of disparate material that I’m trying to achieve. It’s easy to automate a 125bpm techno track to follow another 125bpm techno track, however if I can get the system to blend Frank Ocean into Frank Sinatra without you noticing, then I win.
What are your thoughts on the business of music streaming?
I’ve a lot of friends who make music and, frankly, there’s no way of looking at it without concluding that the whole recorded music scene was irrevocably broken before COVID hit.
Now the live music and events industry is on its knees, the government in the UK is willfully oblivious, and many artists are utterly desperate – either trying to become “content creators” in the hope that they can generate enough traction to feed a Patreon account, or just giving up.
Streaming, from an artists’ perspective, provides minimal benefit, yet we all build our products on the back of their creativity.
Something has to change… but it won’t.
Industry consolidation, combined with streaming and the diminishing of music as a tribal identifier (it’s all games now), means that music has become a ubiquitous commodity that you get unlimited access to as part of your phone contract. How is that supposed to support the next generation of artists? Streaming sucks.
What’s on the horizon for Midtempo Radio?
The stream available at midtempo.net is, in coding terms, an alpha-release. It works, sometimes perfectly, occasionally less so – there’s still considerable room for improvement… Therefore I’m doing what every decent geek would and rewriting the entire platform from scratch.
Deeper analysis and more complex rules means the system will find fewer matches, so the library has to grow (and still maintain some sort of quality control). So, more code, more music, cleverer AI, improved selections, and my constant focus on how elegant (or not) the last transition was.