Nvidia Claims New AI Audio Generator Makes Sounds Never Heard Before (theverge.com) 21
Nvidia has introduced Fugatto, an AI music editor that can generate never-head-of audio combinations, including instruments mimicking animal sounds.
The tool processes both text and audio inputs to create music, sound effects, and modified speech. The system can isolate vocals, swap instruments, and alter voice characteristics.
The tool processes both text and audio inputs to create music, sound effects, and modified speech. The system can isolate vocals, swap instruments, and alter voice characteristics.
How about... (Score:2)
A Norwegian Blue Parrot with 4 million volts through it?
Re: (Score:2)
We should ask it what the fox says.
Re: (Score:2)
Re: (Score:2)
The sound of pining for the fjords.
history repeats (Score:2)
Casio 2024
Re: (Score:2)
A bit earlier than that, I'd wager. It's been a standard feature on kids toy keyboards for ages.
And the tsunami of crap continues (Score:2)
AI has the promise to solve difficult problems and do really useful things
Unfortunately, all the companies seem to release are crap generators
Re: (Score:2)
You want them to go from A to Z in one leap? How fascinating.
Re: (Score:2)
Re: (Score:2)
If you think unheard-of sound generators are crap, don’t go looking at the net worth of a few dubstep “artists” out there. That’ll really piss you off.
Re: (Score:2)
Crap makes money. Diamonds require a more discerning market.
Re: (Score:1)
I agree (and with the other disappointed poster in another thread) - I actually do/teach sound design/music etc. for a living and the demo. in the video seems, to me at least, laughably bad/random/terrible quality, but I can see a lot of management etc types being fooled into making bad decisions (and churning out poor quality work) based on the shameless hype (/their lack of taste), or just wasting everyones time and energy with crap demos.
What particularly irks me is that they have likely ripped off/not
Is this necessarily good? (Score:3)
There are lots of noises I've heard that I never wanted to hear and don't want to hear again. There are some that I do. Many are context sensitive.
But why is "never heard before" automatically desirable?
Re: (Score:2)
Pairs well with "Oooh, Shiny!"
Re: (Score:2)
There are lots of noises I've heard that I never wanted to hear and don't want to hear again. There are some that I do. Many are context sensitive.
But why is "never heard before" automatically desirable?
In my experience, "never heard before" is almost always undesirable. If a particular sound were desirable, someone would likely have created it in the 40-to-60-thousand years that humans have been making music. (Bone flutes actually predate written language, notwithstanding pictographs.)
Re: (Score:2)
And honestly, in the end it's really "just another synthesizer". Because that's what synthesizers are - they're making sounds "never heard before".
Granted, there are many ways synthesizers work from subtractive to additive to FM modulation to wavetable or romplers, but still. Most sounds are just odd and not good, but sometimes you come up with a combination that works, and other times it's become so common it's hard to believe it started with someone
It Also ... (Score:2)
It also makes "sounds I don't want to hear". I listened to the demo - disappointing. The train didn't sound like a train, the lush orchestra was clearly on break, or tuning up their instruments, or butchering a pig - couldn't tell which. The high-pitched chirps were chirpless, and the vocal extraction resulted in some hybrid being that could sing with three mouths. I think they're some distance from a 1.0 release.
Pffffft... (Score:2)
Here's a sound that has never been heard before.
paplay --raw /dev/urandom
Mario Paint (Score:2)
Old news (Score:2)
This is 40 years old, configurable synthesizers were a thing, so many years ago. It obviously never gained popularity and now the sound engineer/band is a laptop in a teenager's bedroom, it still isn't popular.