Song Therapy

• 

  Logo

Inspiration

Lately, I have been seeing, in other people, and have myself, a conundrum, difficulty to find music based on my mood (and who wants to think about choosing music?). Music is a fantastic way to tune out and as Macklemore once said, "Music is therapy. Music moves people. It connects people in ways that no other medium can. It pulls heartstrings. It acts as medicine", but as it is dependent upon our moods, why not make the choice of music automated using artificial intelligence. A fun thing, about mood and music (and a large number of people can attest to this), is that it can increase productivity in menial and boring tasks, music has also been seen to decrease negative feelings of stress and anxiety in people. There is also a special niche of neurologists, that study the effects of music on the brain and its function, called neuromusicology. Research in this area suggests, that listening to music while working can improve your efficiency, creativity, and happiness in terms of work-related tasks, it is also seen that listening to different genres of music can stimulate brain activity in different areas (for example listening to classical and rock helped participants in a 1999 study recognize numbers more quickly). It has also been seen that professionals from all fields benefit from the positive effects of music on the brain (for example surgeons showed better accuracy and efficiency while playing music during surgery and software developers showed more positive moods, better quality of work, and efficiency while listening to music). Using research in this area along with an artificially intelligent system to determine the mood of the user, I created an Alexa skill that determines the mood of the user through talking about anything and based on this mood, recommends a song to help boost productivity and the mental state of the user.

What it does

When a user says to Alexa, "open song therapy" That user will be prompted to talk about anything they desire, but recommended to talk about how their day went. Alexa will then respond by saying, "I sense that you feel ______mood is decided by using AI, would you like to tell me more?" after all conversation between Alexa and the user is finished, Alexa will play a song based on that mood in order to increase productivity and mood of the user. Overall song therapy determines the mood of the user through talking about anything and based on this mood, recommends a song to help boost productivity and the mental state of the user.

How I built it

I built by using a sentiment analysis and natural language processing API called parallel dots, that using LSTM and recurrent neural networks to learn about underlying sentiment in text. I used this API for the sentiment analysis and I put together the database of songs. To program the Alexa I used a lambda function and coded it in python 2.7

Challenges I ran into

Originally, all the music was copyrighted, but the submission crew caught this flaw, and I had to recreate an uncopyrighted database. All the songs now are in the public domain and are free to reuse, I also have the licenses to use all these songs, which anyone can obtain because it's in the public domain. I also originally had trouble with learning how to play music from the Alexa using SSML tags and integrating this in python 2.7

Accomplishments that I'm proud of

I am proud of everything in this skill. I am also proud of the amount I have learned about programming, the human brain, and copyright law in the U.S.

What I learned

I learned a ton about the human brain and how it works, copyright laws in the US and artifically intelligent systems.

What's next for Song Therapy

I may expand the database to more songs and bring it to other platforms such as ios or android.

Built With

• amazon-web-services
• artificial-intelligence
• lambda
• love
• natural-language-processing
• parrellel-dots-api
• python
• the-human-brain