We have different models of generative artificial intelligence on the market, but advances in AI are also going in other directions, some more surprising than others.
However, now researchers have found a way to generate images by AI simply by collecting different sounds from landscapes.
This research has been published in Computers, Environment and Urban Systems, where researchers from the University of Texas at Austin took the characteristic sounds of certain locations around the world in rural and urban environments, and recreated them using artificial intelligence.
That is, this image generator is capable of creating streets only by listening to different audio recordings. To do this, it uses audio and visual data to train.
Computers, Environment and Urban Systems
They demonstrated that the acoustic environments of an area can help represent the visual nature of certain locations.
They previously trained this model with a multitude of YouTube videos and audio clips from cities around the world, from North America, through Europe and ending in Asia.
They were able to create 10-second audio clips and image frames of the locations to train this model, and then compared the images created from 100 audio clips to photos taken of the locations in the real world.
So in the end they created an image generator based on artificial intelligence that relies on sound and is able to capture the scene with precision according to this compendium of acoustic elements.
This is an example of where artificial intelligence can go in the coming decades with results as surprising as this curious image generator.
Get to know how we work in NoticiasVE.
Tags: Artificial intelligence