Scientists have utilized statistical and machine learning methods to predict popular songs with the help of artificial intelligence (AI).
In a study involving 33 participants ranging from 18 to 57 years old, rhythm and PPG heart rhythm sensors were used while they listened to 24 recently released tracks selected by music publishers.
After the auditory experiment, the participants provided feedback on the songs, including whether they found them disturbing, if they had heard them before, and if they would recommend them to others.
By examining the neural responses of the participants to the 24 songs spanning different genres, researchers discovered that AI could effectively predict hit songs using statistical and machine learning methods.
When the data was processed through a linear statistical model, the researchers achieved a 69 per cent accuracy in estimating popularity. However, when the data was evaluated using machine learning techniques, this accuracy increased to 97.2 percent.
Lead author of the study, Professor Paul Zak from Claremont Graduate University, expressed his excitement about the ability of neural activity from just 33 individuals to predict the listening behaviour of millions of people. This level of near-accuracy has not been achieved previously.
The brain signals recorded in the study reflected the brain networks associated with mood and energy levels, suggesting that wearable neuroscience technologies could potentially direct individuals to entertainment that suits their preferences.
Additionally, the study suggested that music companies could use this approach to identify potentially popular songs and include them in listeners' playlists. The data obtained from the participants may also help predict social tastes and preferences.
