Yandex Music’s Nitro neural network analysis technology will help novice authors make recommendations

As the company’s press service informed RB, tests have shown that in one week, Nitro allows young and little-known artists to increase the number of auditions tenfold.

Often, when aspiring artists already have tracks, they are faced with the fact that it is difficult for them to find listeners, because there are many other releases on the Internet and their own audience is too small., Evaluate the potential popularity of your tracks using standard recommendation algorithms. Yandex Music offers a solution to this problem.

Nitro technology uses a neural network to analyze music and compile an audio vector of the composition – a visual image of the track’s sound. It then compares it to audio vectors of other songs or instrumentals.

Due to their similarity to hits, works by young authors can now be included in the service’s recommendations for users. “Nitro” is available specifically for users with an audience of less than 5 thousand listeners per month.

Yandex Music began testing Nitro last fall: since then, more than 600 artists have increased the number of streams by an average of 22 times, when comparing a week using the technology and a week without using it.

The Yandex Music team also plans to compile a rating of the most popular artists based on the growth of Nitro listening, which will be published on the BandLink service for musicians;

To access the new technology, artists must publish a track and submit it for performance using the updated “Pitching in Yandex Music” tool (available on BandLink for verified users).

The Yandex music service will then select the first listeners for each new release. And if they like and listen to the composition, it will end up on Nitro for broadcast to a wider audience.