Machine learning helps researchers identify hits songs with 97% accuracy
Tens of thousands songs are released every day. It is difficult for radio and streaming services to select songs from the constant flow of new releases. These services use artificial intelligence and human listeners to find songs that resonate with a wide audience. The 50% accuracy rate of this approach does not allow it to predict whether a song will be a hit.
Researchers in the US used a machine learning technique that was based on brain responses to predict popular songs with 97% accuracy.
By applying machine learning on neurophysiologic data we were able to almost perfectly identify the hit songs, said Paul Zak. He is a professor from Claremont Graduate University, and senior author of a study published in Frontiers in Artificial intelligence. It is amazing that the neural activity from 33 people could predict whether millions of other people listened to a new song. This level of accuracy has never been seen before.
Source:
https://techxplore.com/news/2023-06-machine-songs-accuracy.html
