Hi there,
So given that its called DJay Pro “AI”, does the software “learn” how to remove elements better as time goes on? Maybe as you listen to more and varied songs or something like that? Hopefully this isn’t a silly question, but I was just wondering. TIA for any replies. L8r!
Yeah - I was wondering wtf the “AI” portion actually means in terms of utilizing Apple silicon? I mean, I primarily use an older Intel-based mac and I honestly see no difference in output - i’m just guessing its a buzzword being used for “Optimized for Apple silicon”, right? I mean, it doesn’t make stems sound better, just might generate them a smidge faster…
SO “AI” doesn’t mean it’s learning, it’s just…better overall performance on Apple Silicon, right?
I’m going to say no, probably not.
A.I pro is also available on Windows so no, it has nothing to do with apple silicone.
“AI” was first added to the djay Pro name back in June 2020 when Algoriddim introduced Neural Mix real-time STEMS separation. See the full press release below. Based on this info, it sounds to me like the AI is being used for very specific tasks and I don’t believe the application is actively learning based on how it’s being used.
"We are thrilled to introduce djay Pro AI and Algoriddim’s revolutionary Neural Mix™️ technology allowing DJs to play and remix singular elements from multiple tracks in real-time. Instantly isolate beats, instruments, and vocals of any song based on cutting-edge AI.
From the analog to the digital age, DJ technology has seen numerous transformations which had a profound impact on music styles and genres as a whole. Neural Mix™️ is a major paradigm shift. Inspired by the auditory system of the human brain, it changes the way we conceive and mix music. It allows artists of all skill levels to experience and interact with music at an unprecedented depth.
djay Pro AI introduces Algoriddim’s brand-new, patent pending¹ Neural Mix™️ technology allowing DJs to isolate, play, and remix singular elements from multiple tracks in real-time. This presents an opportunity to layer and swap individual elements of a track, placing the vocals or rhythm from one track directly onto another, allowing DJs to blur genres and redefine their style.
On screen, djay Pro AI provides intuitive sliders and controls, enabling users to seamlessly fade between percussive and harmonic components of a song, best suited for electronic music, or to create instant acapellas and instrumentals with the tap of a button, best suited for Hip-Hop related music.
Built from the ground up using Core ML, djay Pro AI is the first music app to fully leverage the Apple Bionic chip for real-time playback of music. Neural Mix™️ provides ultra low latency, pristine sound quality, and sample-accurate precision for seeking, scratching, and time-stretching. It does this while processing millions of audio frames per second through deep convolutional networks on the Apple Bionic and rendering high-resolution multi-track waveforms and video in real-time."
Later in December 2020 they introduced the AI powered hand tracking system. Again see the press release below:
"December 09, 2020
We are thrilled to announced a major update to djay Pro AI on iPad, introducing first-of-its-kind gesture-based control of the app, allowing users to mix and manipulate music on a mobile device hands-free.
Users can choose from a set of intuitive hand gestures to scratch, create loops in perfect sync with the beat, and apply filters and effects simultaneously; or simply lean back and perform a clap gesture which triggers an automatic transition from one song to the next. Leveraging the power of Apple’s new A14 Bionic chip on iPad, these hand gestures are intelligently detected in three dimensional space and mapped to advanced DJ performance techniques in real time, with maximal accuracy and instantaneous reaction time."
Finally in May 2022 they introduced AI powered DVS. The only part of this release that kind of mentions AI is the DVS Neural Mix on the b-side of the control vinyl and the “stunning new AI based audio effects that are rendered in real time,”
"May 31, 2022
We are thrilled to announce one of the biggest updates to djay ever – version 4.0 for Mac, iPad, and iPhone – including Digital Vinyl Control (DVS) and countless new features. Take the stage – anywhere, anytime – using your mobile device connected to physical turntables and mixers.
Working closely with some of the world’s top DJs, including award-winning turntablist DJ Angelo, we have also carefully adapted djay’s user interface to perfectly complement any connected hardware and provide a unique blend of tactile and touch based control. Moreover, djay’s audio engine has been fine tuned for digital vinyl control to offer the tightest scratching, most accurate time-stretching, and lowest latency possible.
We are also excited to introduce our new DVS format, Neural Mix™ Vinyl: offering isolated stems for any song right on the B-side of the control vinyl. With the official djay Control Vinyl, you can simply drop the needle of the analog turntable onto one of three distinct tracks on the physical record to play either the original song, its instrumental version, or the acapella.
Other NEW features and updates included in this major update to djay Pro AI:
Advanced DVS options: including Internal, Relative, Absolute modes, Muted/Gated Cue, Seamless calibration, Neural Mix Vinyl settings for B-Side (Instrumental, Acapella vs. Drums, Melodic)
New audio effects: includes brand new Instant FX and a full suite of stunning new AI based audio effects that are rendered in real time"
All three of the examples you Posted only speak to the enhanced performance of Apple Silicon, but there is nothing listed that isnt able to be done on Intel chips (Mac). So it just seems like everything is snappier?
@aaron1138 exactly my point regarding the original post. This is getting a bit off topic, but there is a recent Windows beta release of djay Pro AI. I have run it on my Windows gaming laptop and it performs really well - better than expected actually. However, this PC was considerably more expensive that either my MacBook Air M1 or iPad Air 4. In my experience, Apple hardware is simply better suited for audio and DJ applications. In addition, Algoriddim has up until recently, been primary focused on developing for Apple hardware.
I imagine the developers used machine learning to train the software and this is what the ai is refering to.
Exactly the point I was trying to make above.
Yup.
[stupid20charfiller]
Has anyone tried AudioModify maybe, what are your experiences with it? I found that it has some cool features when it comes to sound, and even a voice customization works really good. Worth a try definitely, you will even get a royalty free output so you can publish.
I think there is some misunderstanding about how an neural networks (NN) work.
There is currently no consumer level NN AI that ‘learns’ on its own that I’m aware of - so the answer to this question is no.
One of the focuses on artificial general intelligence (AGI) is having the NN ‘reinforce’ itself of the fly (i.e. learn). To understand what that means, you need to understand that there are two portions to the progress of NN AI which are relevant - the ‘brain’ itself, and training - this understanding is also relevant to the question of the specialised ‘silicon’ available to AI.
A NN brain consist of a number of objects that are modelled on the human neurone. They have a number of inputs, that are combined mathematically, with a ‘weight’ factor. If the number is above a certain value, the neurone ‘fires’ it’s output. This is basically like the electrical inputs to a biological neurone being resisted by the fatty layer across the nerve.
Training consists of initially randomly attributing a value to weight across all such neurones in an AI brain, and then ‘training’ a large number of these randomly created brains, picking the most successful combinations of the weights, and the 'breeding; those brains, by modifying the weights in relation to eachother (the successful brains) and then repeating the training. Essentially, evolving the brain.
In life, a brain alters itself by modifying the amount of fat on each nerve in response to it’s own success. The beauty of NN’s in technology is that they are super light weight, relative to the complexity of function - I have a friend who routinely messes around with a brain of around 48K cells on his M2 Macbook - but training involves creating many MANY of these brains, and simulating interactons and testing the success of these NNs. It’s very intensive, but once you have it, the NN is very valueable.
Current research is looking for ways to let NN’s efficiently evaluate their own weights based on stimuli, and ‘reinforce’ themselves (like behavioural reinforcement), but it makes the brain much less efficient.
The use of dedicated hardware for AI’s is not a pre-requisite to using an AI - so comment about ‘Windows’ meaning that AI pro has nothing to do with Apple Silicon is not true. Any AI built using current code libraries that are made to work effectively with dedicated AI Hardware will see a performance improvement. This is in much the same way that 3D graphics can be done with just a CPU - but dedicated 3D hardware provides huge performance boosts by creating graphic specific functions at a hardware level.
I hope this helps people understand what’s going on. I have opinions about how Algoriddim could continue to improve with this technology, but I’ve written enough, and I’m sure the Devs are talking about it all day every day 
