Tens of thousands of songs are released every day. This constant stream of options makes it difficult for streaming services and radio stations to choose which songs to add to their playlists. These services use human listeners and artificial intelligence to find what resonates with most audiences. However, this approach is only 50% accurate and cannot reliably predict whether a song will be a hit.
Now, US researchers were able to predict hit songs with 97% accuracy using comprehensive machine learning techniques applied to brain responses.
“By applying machine learning to neurophysiological data, we were able to perfect hit songs,” said Paul Zak, a professor at Claremont Graduate University and the lead author of the study published in Frontiers in Artificial Intelligence. We were able to identify it,” he said. “It is quite amazing that the neural activity of 33 people can predict whether millions of other people have heard a new song. I have never been
Machine learning using neurological data
Study participants were equipped with off-the-shelf sensors, listened to a set of 24 songs, and were asked about their preferences and some demographic data. During the experiment, scientists measured participants’ neurophysiological responses to the song. “The brain signals we collect reflect the activity of brain networks related to mood and energy levels,” Zack said. This allows researchers to predict market outcomes, such as the number of song streams, based on a small amount of data.
This approach is called “neural prediction”. Capture the neural activity of a small number of people to predict population-level impact without measuring the brain activity of hundreds of people.
After data collection, researchers used various statistical approaches to assess the predictive accuracy of neurophysiological variables. This allowed us to directly compare the models. To improve prediction accuracy, we trained an ML model that tested different algorithms to get the best prediction results.
They found that a linear statistical model identified hits with a 69% success rate. After applying machine learning to the collected data, the percentage of correctly identified hits jumped to 97%. They also applied machine learning to his neural reactions in the first minute of the song. In this case, hits were correctly identified with a success rate of 82%.
“This means that streaming services can more efficiently and easily identify new songs that are likely to hit people’s playlists, making their job easier and pleasing their listeners.” explained Zach.
Replication method
“In the future, when wearable neurotechnologies like the one we used for this study become commonplace, it is possible that appropriate entertainment will be sent to the viewer based on their neurophysiology. Instead of being offered , the audience may be given just two or three, making it easier and faster for them to choose the music they enjoy,” said Zack. Told.
Despite his team’s near-perfect prediction results, the researchers noted some limitations. For example, they used a relatively small number of songs in their analysis. Additionally, the demographics of study participants were moderately diverse, but did not include members of specific ethnic or age groups.
Nevertheless, the researchers anticipate that their approach will likely have uses beyond identifying hit songs, due in part to its ease of implementation. “Our main contribution is methodology. This approach could also be used for hit prediction for many other types of entertainment, such as movies and TV shows,” Zak concluded.
/ Open to the public. This material from the original organization/author may be of the nature of its time and has been edited for clarity, style and length. Mirage.News does not take any organizational positions or positions and all views, positions and conclusions expressed herein are those of the authors only. Read the full article here.
