Artificial intelligence produced realistic street views with the help of street sounds

Artificial intelligence Provides realistic street views, although its material is merely the sounds of the same street.

Sound scenery and positioning researchers at Texas University of Austin were training artificial intelligence to create a street landscape.

Professor Yuhao kang He trained learning artificial intelligence with video clips that were compiled from stopped images and sounds in the environment.

Audiovisual clips were about ten seconds long. They were taken from YouTube videos and generally described the prospects of the city and the countryside in North America, Asia and Europe.

SYVÄOPIVIVE The algorithm learned which sounds corresponded to which of the images and the visual environment.

After training, artificial intelligence had to produce pictures itself. It received one hundred sound samples from environments that describe certain street views. Artificial intelligence produced one picture per video.

People were shown these images created by artificial intelligence alongside two other images produced. At the same time, the subjects listened to the soundtrack on which the picture was based.

”

The street video filmed at night was also a night made by artificial intelligence.

When the respondents were asked to identify which of the three images corresponded to the soundtrack, they hit the right eight times out of ten.

“Acoustic environments contain enough visual clues to create an accurate, recognizable street scene,” Kang said based on the experiment In a news release from the University of Texas.

“We can convert acoustic environments into lively visual performances and sounds into sights. When you close your eyes and listen, the sounds around you will paint pictures in your mind, ”Kang says.

“For example, the hum of traffic reminds me of a lively urban landscape. The leaf rustling will take you to a peaceful forest. ”

Kang is an assistant professor of geography and environmental science at the University of Texas and one of the authors of research. The study was published by the science magazine Computers, Environment and Urban Systems.

Artificial intelligence The images produced were investigated in more detail on the computer.

In the pictures, the relationship between the sky, the greenery and the buildings was strongly in line with the original genuine video clips. The pictures even survived the architectural style of the buildings.

Often the images produced by artificial intelligence also really reflected the light that prevailed when taking the picture. So the weather was sunny or cloudy, for example, and the street video filmed at night was also made by artificial intelligence.

At night, artificial intelligence recognized, for example, that traffic noise was reduced at night. The sounds of the insects also change when the evening and darkness come.

Technique can provide applications. For example, a criminal investigator can get an idea of ​​where the sound recording of the crime may have been made.

Artificial intelligence can be used to control urban design and improve the quality of the streets, says researchers at Texas University.

Before this artificial intelligence, the street view and sound had been combined differently, as if in the opposite order.

Artificial intelligence previously succeeded in creating sound effects that corresponded to the view and stored views of a street or other places, says the website New Atlas.