So, it seems that AI is already going to become a thing...that can & will be used to help us, or kill us. Who knows.
But today, its used in making upscaled and better pictures, digital tasks, robotics, military applications, creating or emulating art and creating Deep Fakes and voices of (famous) )people. And yes , its creeping into music in various forms----and of course music creation.
Some applications like Sonible and others already have created apps that are based and programed using machine learning / AI. (I know about the White Sea Studio - Sonible interview). As do you perhaps.
Maybe its not quite AI, just some fancy new expression for different types of machine learning?
But it got me thinking, since well, this is just the beginning of us using and getting used to AI/ML in our life.
I already like the AI app I have on my phone, that can create unique never before seen pictures? Will we see some crazy awesome applications regarding the music and film industry?
Do you guys think that we will be able to use AI in the future to create some crazy unthinkable new apps, synths and plugins?

 

I briefly tried Riffusion, which generates music for your prompts.

It's very hit and miss. I was impressed by some of the seamless transitions it can generate that I'd never have thought of or heard of, like "polka in the style of metallica", but the actual quality of the songs is questionable.

So for the time being, creative AI sits in a place where it's a really cool endless idea and inspiration generator that will definitely break some new ground, but it's not yet a replacement for a real composer or painter for most people.

As for plugins - I don't think we've seen any real AI plugins yet. They're mostly replaying simple learned behaviours depending on the input.

This guy tried getting ChatGPT to write a guitar pedal plugin for him:

Spoiler alert: Didn't quite work out, but pretty close.
Other people have made simple Reaper scripts with ChatGPT that actually work. (Select this track, then disable that plugin..)

I think we're not far away from a mastering AI plugin that will be able to replace a human mastering engineer for most people.

 

I came across this yesterday...

https://gizmodo.com/microsoft-ai-voice-mimic-deepfake-natural-copy-audio-1849969596

 

Its amazing. I remember like 10-12 years ago, this was actually possible with an Adobe beta testing application, i think it was called Adobe Voco...you could record someone, and via text to speech add things the guy never said. As wonderful and amazing as it can be very dangerous....to create shit ppl never did or said...like the thing we call Deep Fake.
Today, as the link you posted, the MS AI is like Adobe Voco on steroids. because back then you had to have like 15-20 min recordings from someone... it was never released to the public, but became "officially known" in 2016-2017...and the alpha and beta versions was available like a few years before that...

EDIT: Adobe Voco in action 6 years ago, public demo:

it was amazing

 

Last edited: Jan 11, 2023

Well, I will do a claim! I will bet in 2025, there will be an site called www.r2r.ai. or ww.vr.ai
The R2R AI or Team VR AI will be available to all, where you upload a .exe, dmg, clap,vst3,dll file, and it will crack it in 5 minutes. And it will learn from itself how to crack the best protections available!

 

AI is already used in movies & documentaries since years (deepfake, old footage restoration, upscaling...)
IRCAM is cloning voices since at least 10 years with machine learning (mostly for documentaries involving dead people), more recently, for french TV https://ircamamplify.com/en/achievements/voice-cloning-for-thierry-ardisson/
AI is already used in mastering services (LandR etc...)
AI is already used in mixing services (RoExAudio etc...)
AI is already used in unmixing services (Demucs, spleeter, UVR and so many others)
AI is already used in the making of many "tradionnal" plugins (sonible, iZotope, Baby Audio, Tone Empire, Guitar ML, ML SoundLab, NeuralDSP...) and in some innovative applications (Google/Magenta DDSP-VST, AI jukebox, Dance Diffusion, Algonaut Atlas, Orb Composer...)

 

Last edited by a moderator: Jan 11, 2023

The future will be amazing regarding AI.

 

Actually the future always will get worse... that's what I've noticed being alive for 3 decades. I miss the days of physical things just working for longer than 5 years. My fucking toilet is 5 years old and is shitting the bed. If Ai could be incorporated into manufacturing things that last longer than 5 years, it would be something I could get behind.

I am personally waiting for an ai helmet or headphones I can wear that can "listen to my thoughts" and turn my chords and melodies in my head into midi while I tap a tempo........ So I may be waiting forever