Travel is super special to everyone! There are moments when you look outside the window and within seconds you are transcended into a totally different world. Some people would be excited and looking forward to their destination while a few others would be totally sentimental. It was during the long bus travel from USC to Stanford that we realized how important songs are to keep the mood up. Being super excited, wouldn't it be intuitive to listen to a lot of happy and energetic songs? When we are emotional it is natural to prefer songs that have a touch of melancholy. This realization inspired us to develop a one-stop solution to play songs from the user's own playlist based on the mood.

What it does

Given a user's playlist, our app "Sound Therapy" suggests and plays songs based on what the type of user's mood is.

How we built it

We take the user's audio files, transform them into images. The next step is to train these cool images using Google's AutoML to classify the songs based on mood!

Challenges we ran into

Coming up with appropriately formatted datasets and visualizing audio frequency using WebAudio API

Accomplishments that we're proud of

We were able to train the model at the lowest cost and with lesser elapsed time!

What we learned

A lot of low-level math behind images and sounds!

What's next for Sound classifier

Integrate with the top music apps like Spotify, Youtube, etc to categorize the user's playlist

Built With

Share this project: