Brain Data Analysis Predicts Chart Hit Songs With Near-Perfect Accuracy

Anticipating chart-topping songs has always been challenging. However, scientists have recently leveraged machine learning (ML) coupled with high-frequency neurophysiological data to significantly enhance the accuracy of hit song predictions. This cutting-edge combination demonstrated near-perfect accuracy in predicting hit songs based on neural responses of individuals while they were listening to new music. The next generation of pop tracks may arrive optimized for the brain.

Picking the best beats

Streaming services and radio stations grapple daily with a deluge of new songs. Sifting through these tracks to build playlists is an involved and time-consuming mission. Traditional methods to identify crowd-pleasing songs have included relying on human listeners and artificial intelligence. But even these techniques have little to show for themselves – racking up a hit-predicting accuracy rate of less than 50%.

In the new study, a team of U.S.-based researchers has blown that mark out of the water by employing a comprehensive machine learning technique for brain responses, achieving a remarkable hit song prediction accuracy rate of 97%.

“By applying machine learning to neurophysiologic data, we could almost perfectly identify hit songs. That the neural activity of 33 people can predict if millions of others listened to new songs is quite amazing. Nothing close to this accuracy has ever been shown before,” said Paul Zak, a professor at Claremont Graduate University and senior author of the study published in Frontiers in Artificial Intelligence.

Forecasting from brain data

During the study, participants – outfitted with a commercially available sensor platform – listened to a 24-song playlist. Their neurophysiological responses were measured throughout. The recorded data reflects activation of a brain network linked to energy levels and mood, said Zak. This novel technique, known as ”neuroforecasting”, harnesses neural activity from a limited sample size to forecast large-scale effects without the requirement to monitor the brain activity of hundreds of individuals.

The researchers then applied a battery of statistical tests to investigate how accurate their neurophysiological predictions were. These predictions were combined with several ML models to assess which approach produced the most accurate outcome. After these techniques were applied, the song identification rate soared from 69% to 97%. Even when brain activity from the first minute of the songs was used, accuracy stayed high at 82%.

Freedom of choice?

“This means that streaming services can readily identify new songs that are likely to be hits for people’s playlists more efficiently, making the streaming services’ jobs easier and delighting listeners,” Zak explained.

He added, “If in the future wearable neuroscience technologies, like the ones we used for this study, become commonplace, the right entertainment could be sent to audiences based on their neurophysiology. Instead of being offered hundreds of choices, they might be given just two or three, making it easier and faster for them to choose music that they will enjoy.”

The accuracy of the team’s results belies some important limitations in the work. They used only a few songs, and certain age groups and ethnicities were not included in the analysis. Zak remained optimistic about the technology’s use cases, which he believes could extend beyond deciding what is top of the pops: “Our key contribution is the methodology. It is likely that this approach can be used to predict hits for many other kinds of entertainment too, including movies and TV shows.”

Reference: Merritt SH, Gaffuri K, Zak PJ. Accurately predicting hit songs using neurophysiology and machine learning. Front. Artif. Intell. 2023;6. doi: 10.3389/frai.2023.1154663

This article is a rework of a press release issued by Frontiers. Material has been edited for length and content.