GabyThe era of artificial intelligence (AI) has already arrived, and over time, neural networks will become even smarter. Therefore, it can already be stated that their impact on art and pop culture will be significant. There are already convincing deep fakes that are causing panic among recording brands.
However, one should not dismiss the potential benefits of AI in the music industry. One of the likely advantages of artificial intelligence for some studio executives is the ability to predict which musical compositions could become undeniable hits.
How exactly do neural networks work? Will they be able to determine future hits?
Researchers used the neural activity of 33 people who listened to 24. songs , in combination with statistical modeling. As a result, the neural network was able to almost perfectly predict which composition would become a true hit and which would not succeed.
The lead author of the study, Professor Paul Zak, claims that by giving artificial intelligence access to neurophysiological data, his team was able to almost perfectly predict hits. In his opinion, this is an incredible breakthrough, as the neural network was able to predict whether millions of other listeners would be thrilled by new songs based on the neural activity of just 33 participants in the experiment. According to the professor, nothing close to such accuracy has ever been observed before.
Participants in the study, aged 18 to 57, were equipped with heart rate sensors and a music player containing 24 recently released songs. The tracks were selected by the staff of the music service. A song was considered a “hit” if it received over 700,000 streams. The playlist included a variety of music genres: 13 hits and 11 unsuccessful tracks.
After the listening experiment, participants completed a survey about the songs. The survey included questions such as whether the composition was offensive, if they had heard the song before, and if they could recommend it to their friends.
However, the main role in the research was played by the neurophysiological reaction to the songs. Thus, researchers were able to predict the reactions of other people—whether a particular composition would be successful or a failure—through “neuroforecasting” based on a small dataset from 33 participants. To achieve accurate results, they did not need to verify data from thousands of listeners.
The neural network provided an incredibly accurate forecast for millions, based on data from a small group of respondents. As Professor Zak notes, the collected brain signals reflect the activity of the brain network associated with mood and energy levels.
Researchers found that when processing data using a linear statistical model, the success rate of predictions was 69%. This was not critical, but artificial intelligence achieved results with an accuracy of 97.2% when performing the same task. Scientists note that even when the neural network evaluated data based on just 1 minute of listening to a song, the accuracy was still 82%.
Forecasting hits not only in music
Mr. Zak says that if similar neuroscience technologies become commonplace in the future, suitable entertainment could be offered to audiences based on their neurophysiology. Instead of presenting listeners with hundreds of options, they could be provided with just 2-3 compositions, making it easier and faster for them to choose music they will enjoy.
Researchers focused on their work, assuming that music companies could use it for “easily identifying new songs as potential hits that people will add to their playlists.”
Although the research had limitations due to the narrow selection of songs and audience, it is not hard to imagine a future where music, television, and film do not go through a demonstration stage. Especially if AI models can predict success among users with an accuracy better than 80% after just one minute of media usage.
However, since approximately 100,000 new songs are uploaded to the internet every day, it cannot be said that music fans’ choices will become limited anytime soon. However, as noted by the publication Newatlas Professor Zak adds that “it is quite likely that this approach can be used to predict hits for many other forms of entertainment, including movies and TV shows.”
