- Joined
- Feb 24, 2020
A thought comes to mind...
Would a similar tool one day be used to generate AI music? Like not just generating random sounds to a tempo, but done in a way that you would really struggle to tell the difference? Maybe even to the point where it generates its own samples and uses them in a similar fashion as a musician would to generate music that follows certain musical rules?
I know it has been done before, with uh, varying results (mostly shit). And as for voice synthesizing, I distinctly remember some sort of Adobe program being showcased in which it could basically replace words spoken by feeding it a sample, pausing where you want to change the wording and simply writing down the new words... Pretty neat, if it was legit... But either way, I can see voice synthesizing being much easier to reproduce than straight up entire musical scores.
And for the Hand Question, yea it's actually really difficult to draw hands, but the AI is struggling with much more than that, it's the fine details that have to be enchanced a bunch of times over, but if it doesn't really understand how joint structure works for didgits and limbs, with hands or feet or even legs some times (I've seen pictures where the legs contort in really weird ways), then it of course it can't really perfect something it doesn't fundementally know how to create. I suppose it's rather poetic really, artists use their hands to draw, they see them all the time, and with practice they can improve. And this AI is pure math, trying to desperately figure out what the hell these thin sticky points are that stretch out of the arm noodles.
Loooong ways to go yet.
Would a similar tool one day be used to generate AI music? Like not just generating random sounds to a tempo, but done in a way that you would really struggle to tell the difference? Maybe even to the point where it generates its own samples and uses them in a similar fashion as a musician would to generate music that follows certain musical rules?
I know it has been done before, with uh, varying results (mostly shit). And as for voice synthesizing, I distinctly remember some sort of Adobe program being showcased in which it could basically replace words spoken by feeding it a sample, pausing where you want to change the wording and simply writing down the new words... Pretty neat, if it was legit... But either way, I can see voice synthesizing being much easier to reproduce than straight up entire musical scores.
And for the Hand Question, yea it's actually really difficult to draw hands, but the AI is struggling with much more than that, it's the fine details that have to be enchanced a bunch of times over, but if it doesn't really understand how joint structure works for didgits and limbs, with hands or feet or even legs some times (I've seen pictures where the legs contort in really weird ways), then it of course it can't really perfect something it doesn't fundementally know how to create. I suppose it's rather poetic really, artists use their hands to draw, they see them all the time, and with practice they can improve. And this AI is pure math, trying to desperately figure out what the hell these thin sticky points are that stretch out of the arm noodles.
Loooong ways to go yet.
Last edited: