Machine learning today not only recommends the things you can buy, or content you can watch, but it is doing wonders in other domains as well. This time it is on a mission to find out something untouched. There are different elements in music that trigger emotion in humans. And machine learning is trying to find out just that — how music affects brain activity, physiological response, and human-reported behaviour.
Music & Emotions
New research by scholars from the University of Southern California is trying to figure out the elements in a song that triggers different emotions in a human. To be more specific, the project has been carried out considering things like dynamics, timbre, harmony, rhythm, and register. They are also trying to figure out how machine learning could use elements and relate to emotions to predict how people might respond to a new piece of music.
“This work adds to our understanding of how music affects multimodal human experience and has applications in affective computing, music emotion recognition, neuroscience, and music information retrieval,” researchers stated in their paper.
This is the process that the researchers used:
Using Human Efforts
- They first gathered songs from music streaming sites considering that they have very few plays, and are either tagged with “happy” or “sad”
- Then they sorted the songs (60 pieces for each emotion) and it was done with the help of
- Furthermore, they created three more groups where two groups had triggers sadness and one group had only songs that induced happiness.
- The next thing they did is invited 100 people to listen to those songs from the three groups. And once they were done, they had to fMRI scan. They even had to wear sensors to track pulse, heat, and electricity in order to rate the emotions.
Using Machine Learning
- Once all the data was collected by human testers, they fed that data to a machine learning algorithm.
- That is not all, along with that data, they also fed the machine with features of a song such as pitch, rhythm, harmony, etc.
The early results of this project has shown how in the future machine learning would be able to do so much more in the music and art domain. Despite the fact that it is already doing wonders, the researchers are saying there is still a lot to explore. The project is still in the early stage and it would take a little more time to come to a solid conclusion.
However, they are also optimistic that once it is done and successful, machine learning would not only help you select your bedtime or gym playlist but would also help movie makers, therapists etc. to come up with targeted musical experiences.
This is not the first time when machine learning has been used in the music industry, in recent times we have seen machine learning making heavy metal music, painting portraits, and even beatboxing. And with time this sought after tech is just making things easier for the human race.
Despite all this, there are people who would argue that these advancements of technology would soon take up all the jobs of the artists. There would be a time when making music would be so much easier, that the need to have an artist won’t be there. But that’s not the complete story, there are professionals from the industry who believe that these technologies would only empower them to be the best version of themselves. It would make some of the complicated things easier for artists.