Artificial Intelligence To Feel Emotions In Music
The app market has several music applications like Spotify, iTunes, Google Play Music, YouTube Music, etc. that allow people to stream and download music with a minimal subscription. Most of us are all well acquainted with these apps as we listen to our favorite music on them regularly.
Music apps have a common feature that categorizes music as per the mood and emotions. For example, Happy Song Playlist, Sad Songs Playlist and likewise. But oftentimes you tune in on a playlist to liven up the mood but it turns out that not every song in the list can be put under the happy category.
Sometimes it turns out to be a random compilation of the latest songs. Hopefully, the present flaw in the categorization of songs emotion wise can be improved. In this case, the technology that comes in handy is Artificial Intelligence.
Musical Experiment with AI:
A group of researchers at The University of Southern California (USC) is conducting a study that uses AI-based tools like Machine Learning to read the emotions in music. According to them, music has elements like rhythm and tones that induce emotional reactions from the body.
Music instigates a wide array of brain activity in the form of physiological reactions and emotions such as happiness, sadness, melancholy, etc. The researchers are of this view that artificial intelligence can read the beats and that way predict how the listener will react to music emotionally.
The main subject of the experiment is to understand how any kind of media such as movies especially music affects the human body and mind and they use AI for it.
Shrikanth Narayanan, professor of USC elaborated on this experiment, “Once we understand how the media can affect your various emotions, then we can try to use it productively to support or improve human experiences.
The researchers tested the theory with around sixty songs for each human emotion. They concluded a final list of three, out of which two songs Ólafur Arnalds’ “Fyrsta” and Michael Kamen’s “Discovery of the Camp” induced sadness. Lullatone’s “Race against the Sunset” was the one that induced happiness.
In the second phase of the experiment, around one hundred participants who had never heard the songs before were divided into two separate groups. The participants listened to the three songs and took an exploration of heat, pulse and electricity sensors on their skin. The intensity of their emotions got rated on a scale of 0 to 10, which provided the data for the AI to learn and take note of.
Results of the Experiment:
When the researchers input the collected data into multiple AI-based algorithms and further examined what features were strongest in predicting the response. It was found that the level of their medium and high frequencies and the strength of their rhythm were among the best predictors.
The research is still in the preliminary stage, however, researchers are quite excited about the application of these models. Some of the examples that they furnished were, creating music for specific people, maybe to help ones with mental health issues. They have indicated that AI is a powerful tool for knowing human intelligence itself. Such a system can bring significant change to the present music industry and it makes as well.