Are Computers Going To Completely Replace DJs?

Matthew Owen Reininger | Read time: 5 mins
future djing
Last updated 26 November, 2017


Industrial Robot DJ
Are computers about to make human DJ’s history? And will the dancefloor even notice?

Is the DJ dead, or at least, dying? A bold question for sure, but with dramatic changes taking place in both digital DJ technology and data analytics on the internet, this could be just what’s happening all around us, right now. Could the role we all love really be edging closer to being replaced by our digital DJ sidekick, the computer?

Looking at the state of play in music technology, social media, and data analytics, it’s possible to catch a glimpse of where we might be heading in this brave new world of 21st century DJing – and you may not like what it looks like…

Is all this new technology really our friend?

It certainly appears that today, technology is advancing faster than ever. Many people argue that software is replicating the technical and artful aspects of DJing, and personally I believe that the software is closing in on doing most of that work really rather well.

Is this such a big deal? Many digital DJs would believe it not to be. It seems that the consensus among supporters of digital DJing is that the technical aspects of DJing – beatmatching, mixing, beat juggling/looping, blending and so on – have already been partially or fully outsourced to our computers, and that this is just fine.

If an algorithm can do the process for us, goes the argument, it allows us to focus on better selection of music and different creative performance elements (eg effects, remixing…). And to take the argument to a deeper artistic level – as many have argued endlessly in digital DJing forums across the internet – if technology can now perform these actions for us, how much of a skill was it really in the first place?

Could computers learn music programming?

But here’s where the argument takes a sharp turn. The most coveted DJ skill for many DJs is music selection and programming. Delegating this responsibility to a computer is a sacred cow in the DJ community (some of you may be thinking about the TouchTunes jukebox at your local bar right now…) .

Mood Knobs was shown off at the London Music Hack Day 2011. Developed by Alex Michael, it is a Spotify hack that allows the user to navigate millions of tunes by precisely telling the app about various aspects of their current mood.

The selection and programming of music, goes the argument, is the most human element of what we do as traditional DJs, and as this skill could never be replaced by a computer, we’re all safe. But is that really true? As computers “learn” about our musical tastes, could a computer program perform for a crowd just as well as a flesh-and-blood human DJ?

Consider this fact: Digital music platforms such as Rdio, Spotify, Soundcloud, iTunes Genius, or social music platforms like are great evidence of the power data plays in the way we listen to music today.

These services collect data on what we listen to, and based on our input (“liking” or “disliking” a track) these platforms build stations based on tags such as genre, and build relationships to other similar artists. Some say we are essentially “teaching” the machine about what type of music we want to hear and how (or in what sequence) we want to listen to it in.

Computers can get better at DJing, just like humans

This process is being refined with every mouse / trackpad click. Pandora is an example of the powerful role sociability and personalisation can play in the way we listen, actively, to music.

I remember listening to Pandora around 2008, and it was not nearly as entertaining to listen to as it is today. This has to do with my input over time.

If computers can “learn” to improve upon their already existing algorithms, then I do not see much difference between a novice DJ learning how to program for a crowd, and algorithm that chooses music based on previous choices. Eventually the novice DJ, if playing for crowds, will most likely improve. On a computer, with every like and dislike that a user submits to a music platform, the music choice algorithm improves.

Now, you may be saying: “Well there is a difference between a computer selecting for an individual based on their own personal preferences and a DJ selecting based on how he or she is reading the mood of the crowd, right?”

Why DJing can be as much luck as judgement

Yes there is a difference, but the selection method when reading a crowd is entirely a gamble.

DJ crowd
Can he really claim to know what each and everyone of those people want to hear next?

You have to be able to make decisions based off subjective data: what the crowd danced to for instance, or remembering the tracks you played before, likewise remembering what the crowd enjoyed, as well as what they didn’t, and then being able to plan a musical journey from there, and so on.

This is unreliable because of the possibility of human error. You cannot truly know what the crowd wants to hear without asking them directly.

However, what would happen to a DJ’s set if he or she had real-time data from the audience, sent by club patrons in the same way we like or dislike tracks on our Pandora stations? This way we know what would entertain the crowd with some degree of accuracy.

I for one would trust this as an indicator about what my crowd would like to hear more than say, occasional requests by the drunk lads and ladies prying their way into the booth!

Is this the dancefloor of the future?

Imagine a crowd collectively providing data by liking or disliking a track the “DJ” is playing, making requests from kiosks in the club or from their mobile devices.

This would be too much data for a human DJ to handle (especially after a few drinks!). The clubbers are telling the DJ, in real-time, whether they approve or disapprove of the song selection and mix. The DJ uses this feedback and compares all other data points against the current track selection and adjusts its next track selection accordingly.

Phone clubber
Instant feedback to a computer DJ: Is this how it’s going to be? Pic: Sephie Bergerson

Sounds like something a computer could do? Absolutely!

Theoretically, the computer DJ could simultaneously match genre, tempo, timbre, date of recording, and any number of data tags tied to the track that is currently playing, and then, comparing this data to other tracks in its catalogue, pick an ever-more “perfect” song to mix next.

On the human tip, this sort of involvement from attendees strikes at the essence of what some feel we do as DJs nowadays; being the central figure/concept in DJ culture.

Might this all take us back to the future?

But before the culture entered the superstar DJ era, we were just thought of as the leaders of a party. I remember hearing stories from my mentors about how they would love to, “give a party” to people.

A part of that “giving” aspect is performing for your crowd – knowing the state of music culture at a given time and creating a musical story or soundtrack, reflecting the time and space, for the people involved on the dancefloor. It seems to me that the people used to be a more active part in the process back then then superstar DJs often are now. With newer technology allowing the crowd to democratically program, we might ironically end up getting back to the roots of DJ culture in which the community mattered as much as the DJ at an event. This would be a big cultural deal.

Ray Kurzweil, noted futurist and famed studio sampler engineer, predicted that man will merge with machine by the year 2045.

While how profoundly and how fast computers and people may merge is a bigger argument, it no longer seems such a massive leap to me that the last great bastion of our craft as DJs – music selection – could be automated. And pretty soon at that…

• Matthew Owen Reininger (DJ CNTRL) is an educational technologist and has been a DJ/turntablist for over ten years. He currently calls Circuitree Records home. Thanks to Rob Dyson, producer and engineer at Wizkid Sound in Atlanta, GA for contributing to this article.

What do you think? Is truly automated DJ software, with some kind of audience feedback, just around the corner? How well do your internet tastemaker streams, web radio channels and social streaming service predict your musical wants and needs? Or will there always be the need for a human to be creative in a way computers could never replicate? Over to you…

Click here for your free DJ Gear and software guide