AI Music and VOCAL software and plugins

Discussion in 'Software' started by AgA, Nov 29, 2024 at 3:08 PM.

Tags:
  1. AgA

    AgA Newbie

    Joined:
    Nov 6, 2013
    Messages:
    43
    Likes Received:
    2
    Why haven't the big music companies like UA, Eventide, Softube, waves.... created AI software and plugins yet? and in the same time some new and small companies they did?
     
  2.  
  3. Radio

    Radio Audiosexual

    Joined:
    Sep 20, 2024
    Messages:
    1,012
    Likes Received:
    532
    Hello @AgA, the company " Waves " already uses AI:

    Waves introduces the new Curves Spectral Series, three AI-based EQ plugins to simplify mixing work in the studio.


    AIformusic: 50 tech companies sign new guidelines for AI in music

    How important is artificial intelligence's role in modern production workflows?
    www.gearnews.com/aiformusic-50-tech-companies/

     
    Last edited: Nov 29, 2024 at 3:45 PM
  4. xorome

    xorome Audiosexual

    Joined:
    Sep 28, 2021
    Messages:
    1,170
    Likes Received:
    857
    Waves has a range (~8-10) of plugins claiming to use AI in some form. But since there's no mandatory minimum definition for what constitutes AI, anyone is free to slap "AI" on their product's name - irrespective of what the plugin actually does, including: nothing AI at all. And that's pretty much what bad-faith plugin developers have been doing heavily for the past two years - "Super AI Ultra Compressor AI v2... with AI".

    Bigger companies are by definition slower to react to changing environments than tiny one-man shops with a passion for tech. I'm sure there's at least one guy at every big company scanning for popular hobbyist projects on github at least once every other month to see if there's anything AI-related worth lifting and staking their brand name on.

    That being said, we already have many "AI-informed" plugins, that perform no AI by themselves, but where AI was trained on material and gave informed development directions based on that.

    With current hardware, larger AI models are simply too demanding to run in real time, so the domain has to be tiny to train a tiny model that works at acceptable speeds. (Like WaveNet and RTNeural as used by NAM, ToneX, AIDA-X and dozens others to mimic distortion hardware)
     
Loading...
Loading...