It is now possible to separate vocals, instruments and drums in real time in your DJ sets from any track you are playing, thanks to a new feature called Neural Mix that has just been added to Algoriddim’s djay for IOS app.
Algoriddim has added a feature it is calling “Neural Mix” that separates the acapella from the vocal in any track, and there are also three new faders available per deck, that separate the drums, “harmonics” (music, basically) and vocals, if you want even more options.
Read this next: Neural Mix Pro Brings Standalone Real-Time Stem Separation
This presents an opportunity to layer and swap individual elements of a track, for instance placing the vocals or rhythm from one track directly onto another, allowing DJs to create instant acapellas, mix more creatively, and create live mashups with ease.
The waveforms of the tracks in the software adapt in real time to show you what you’re actually hearing, which is awesome, and as well as faders for the parts, you get mute, solo and a “swap” function that instantly swaps the chosen element with the same one on a track playing on the other decks, for immediate mashups.
Because djay has a built-in sequencer and sample packs available from within the app, it’s simple to, say, isolate a vocal and then build a drum track around it – and as this feature works across all decks, you can get incredibly complex when mixing – almost more like four-track production than DJing – especially because all the effects, key sync and so on available in real time too.
Mapped to DJ hardware
The app comes with mappings for 30 pieces of DJ hardware, and Algoriddim has improved its automix feature too, to take advantage of these new abilities for smoother AI mixing when the DJ isn’t present.
Learn to DJ creatively with Digital DJ Tips: The Complete DJ Course
The feature is currently only for on the iOS version of djay, and while djay for iOS is compatible with any iPhone or iPad running iOS 12 or later, an iOS device with an A12 Bionic chip or later is recommended when using Neural Mix. This includes iPhone XS or later, iPhone SE (2nd generation or later), iPad Pro 11‐inch, iPad Pro 12.9‐inch (3rd generation or later), iPad Air (3rd generation or later), and iPad mini (5th generation or later).
• djay for iOS is a free download on the App Store. Users can upgrade now to djay Pro AI, including access to Neural Mix
First thoughts
So how does it sound? It’s pretty good – certainly good enough for DJing with. Yes, there are artefacts and a bit of bleeding between channels, and the separated parts have got that kind of pumping, slightly artificial quality you hear when somebody has got aggressive noise cancellation on a Zoom call, but it’s definitely usable.
The vocal separation especially is impressive.
You’re never going to want to lay down the output form this kind of AI in your studio productions, but that’s not the point here – the point is that on the spur of the moment, in front of a live audience, you can do stuff that until now has been impossible.
Mobile DJs can drop the vocals out to let their crowds sing along (yes, it works fine on Bon Jovi’s “Livin’ On a Prayer” – we tested it!), house DJ scan drop the drums out for an “instant breakdown”, pop DJs can hit “loop” and kill the vocals on today’s two-minute pop songs to given themselves a bed for beatmixing. And so on.
Algoriddim tells us that this has all been made possible thanks to the AI power of the A12Z Bionic chip in modern iOS devices, which is the reason why it is only available on modern iOS devices for now.
Acapellas come full circle
This will be THE feature other companies rush to catch up with, because it brings us full circle on something started by DJs using acapellas way back in the day, played from vinyl.
Next, Native Instruments released the Stems format for MP3s, allowing DJs to mix with four elements of tracks. But the feature’s downfall was that record labels and producers had to provide versions of the tracks that would work with the feature. Few did.
Programs like XTRAX STEMS then arrived, allowing DJs to extract the drums, vocals and music from tracks themselves, which could be exported to use either in Traktor with the Native Instruments STEMS format, or just as individual files to play in DJ sets.
Now, djay Pro AI can be seen as real-time XTRAX STEMS – or alternatively, Native Instruments’ Stems but working on any track at all you load up.
(One difference is that djay Pro AI only separates drums, music and vocals, whereas STEMS specifies bass and drums separately, but it’s otherwise the same.)
What will this mean for DJing?
What this will mean for DJing (Algoriddim modestly claims it has “reinvented DJing” with the feature) is another thing. While in unskilled hands we can expect more terrible mashups, out of context, out of key, and off beat, clearly in skilled hands it has the potential to be a killer feature.
From mobile DJs dropping out vocals to let the crowd sing along, to house DJs killing drums to make instant breakdowns, the potential is endless and frankly, on first play, pretty mind boggling.
Learn to DJ creatively with Digital DJ Tips: The Complete DJ Course
It will definitely be fun to see how quickly the other software companies catch up (if indeed they do), and how alternative implementations of this idea vary.
And from Algoriddim’s point of view, when djay lost its (pretty-much exclusive) Spotify integration recently, many thought the company would hit hard times – but this feature has clearly put it right back on the map.
We say it a lot, but – exciting times!
Interview with Algoriddim’s CEO, Karim Morsy, about this tech
I spoke to Karim Morsy, the CEO of Algoriddim, about this tech, and put to him many of the questions being asked by the community. Watch the replay here.
What do you think if this new feature? Let us know in the comments.
Phil Morse
