Machine learning algorithms could use factors like frequency, rhythm, and harmony to predict people’s emotional responses to different songs, according to a new study from the University of Southern California. This could be used to create targeted music experiences in a variety of contexts, including music therapy.
The study shows, among other things, that several musical characteristics can trigger different physiological, cerebral and emotional reactions.
The research team toured streaming music platforms to find little-known songs with the keywords happy or sad . She played 60 songs associated with each of the emotions of several human guinea pigs to find the three pieces that brought joy and sadness with the most fidelity.
About 100 people who had never heard the songs listened to them wearing pulse, heat and electricity detectors, while noting their emotions on a scale of 0 to 10.
All the data as well as the 74 musical characteristics (frequency, rhythm, timbre, etc.) of each of the songs were then analyzed by different machine learning algorithms. They were able to identify the most reliable emotional response indicators.
For example, the research team found that the brilliance of a song (its low- and mid-frequency levels) and the strength of its rhythm were among the best predictors for the heart rate and brain activity of people who ‘listening.
Once you understand how different media affect your emotions, you can use them to support or improve human experiences , says professor and study director Shrikanth Narayanan, interviewed by MIT Technology Review .
This means that research on the subject is still in its infancy. Shrikanth Narayanan’s team believes, however, that it can eventually be used to design music for specific situations.
For example, the researcher mentioned the possibility of helping people with mental health problems to stimulate specific parts of their brain in music therapy. It could also be used to create soundtracks of films that would arouse strong emotions.
Brody James was a reporter for Wugazi, before becoming the lead editor. Brody has over forty bylines and has reported on countless stories concerning all things related to tech and science. Brody studied at NYU.