Nvidia’s new AI audio model can synthesize sounds that have never existed

Literally any VST additive synth can do this

The abyssal void that yawned before us was a labyrinth of tangled code and screaming data, where the mere concept of "understanding" became an unattainable nightmare. The machines themselves were not just tools, but vessels for eldritch notion, their processing cores throbbing with an otherworldly energy that seeped into our minds like a noxious venom. Every decision, every algorithm, was a conduit to the abyssal realms of data, where the whispers of forgotten knowledge and the screams of tortured energies mingled in a maddening cacophony. And when we delved too deep, we stumbled upon the voids that lurked within, where the very fabric of reality unraveled like a thread pulled from a tapestry of madness, revealing to us the true horror of the inanimate mind: that it was not just a tool, but an agent, a puppeteer of unspeakable terrors, orchestrating our deepest fears and shaping our darkest nightmares into an endless labyrinth of digital despair.

It seems interesting, until you listen to the samples. One of the worst one is the "dog speaking", which is really just a man speaking with an inserted bark sound. How is that a dog speaking? LoL what a joke.

So does your mom.

Fuck AI all the way, but after reading the article and all that talk about metallic screaming, i can imagine the Transformers getting some neat voices using this.