Music Therapy with AI

Inspiration

The inspiration came from a friend of mine. The second Artsakh War left behind many boys and men with awful cases of PTSD. Many institutions helped them; however, none of them had music therapy. Because of that, my friend decided to leave her high-paying job in a bank and go to Germany to study music therapy. After finishing her studies, she will return and use what she has learned to help the soldiers.

When I was thinking about a project to do, she came to mind, and I decided to base my project on music therapy.

What it does

One of the principles of music therapy is that when you're sad, you should listen to sad music, and when you're happy, you should listen to a happy one. Music therapy with AI does precisely that; it detects your mood and plays suitable music.

Therapists can also use this program during therapy sessions. During the sessions, patients are encouraged to talk about their experiences, thoughts, and feelings. The patient will face the camera, and when AI detects signs of distress, it will play calming music.

And you can always use it as a playlist controller, changing the tracks with your facial expressions:)

Challenges we ran into

The original idea was to use an accelerometer to detect signs of distress based on hand movements. However, accelerometer data is a less researched field, and it was hard to make it work. That's why I decided to use facial expression recognition instead of accelerometer data, and I think it turned out better:)

What's next for Music Therapy with AI

In the future, I plan to add a function of adding more classes for other emotions. I'd also like to add the option of choosing youtube playlists as your tracks.