Mobile operator Tele2 and big data business solutions developer Platforma used a neural network to create city music. Artificial intelligence generated five sound compositions reflecting the sound of Moscow, St. Petersburg, Nizhny Novgorod, Arkhangelsk and Kazan. The tracks were written based on the city’s skyline and geodata, including building density and regional traffic.

Image Source: fitzgeraldbc/pixabay.com

Artificial intelligence transformed the horizon line into a wave that determined the timbre of a particular composition, geodata on population density were converted into a sequencer that “arranged notes in place and in time.” The music written by the neural network was adapted by popular artists living in the cities whose data was processed. The tracks can be found here .

According to Anton Merzlyakov, director of big data analytics at Tele2, the musical project is designed to show that neural networks are equally good at displaying a special city code in notes and identifying places of implicit tourist attraction.

“Many people associate cities with a certain smell or image, but we decided to generate their unique sounds. With the help of big data, we took into account a variety of indicators: from the number of tourists to building density and skyline, thanks to which we got a unique and unlike anything else sound. We have chosen five cities – Arkhangelsk, Moscow, St. Petersburg, Nizhny Novgorod and Kazan – as they most clearly demonstrate Russia’s landscape diversity and rich architectural heritage. And the musicians, the indigenous inhabitants of these cities, helped to display it in a language understandable to everyone in the world – the language of music,” Platforma said in a statement.