The researchers team has developed a new model of artificial intelligence called Fugatto, which is capable of generating sounds that have never been heard before. The development carried out by NVIDIA specializes in the creation of audio from text prompts and is, according to the developers, a kind of "Swiss army knife" for sounds. The model is capable of creating unique sound compositions that can be used to edit, as well as to generate new sounds that have previously been impossible.
As Richard Kerris explained by Nvidia, Fugatto is a more agile and versatile model than other similar technologies. It is able to generate sounds such as the "barking of the pipe" or "saxophone". For example, on the basis of a text clue "deep, roar bass pulses paired with intermittent, high -frequency digital chirping, as the sound of a waking up massive smart machine", the model created a new unique sound that could not be reproduced by ordinary tools.
In addition, Fugatto is able to transform sounds from one type to another, for example, converting the sound of the train into a string orchestra. The producer ido mixed, who participates in the Nvidia Inception program, said: "This thing is wild. Sound is my inspiration. The idea that I can create brand new sounds on the fly in the studio is incredible."
Fugatto's development has taken over a year and has used millions of audio-cubs to study. However, despite its potential in music and sound design, there is fears of the impact of this technology on traditional creators and their art.
