Fortress

Music recommendations based on mood inputs

Inspiration

During the early stages of the hackathon, we were listening to music as a group, and were trying to select some songs. However, we were all in slightly different moods and so were feeling different kinds of music. We realized that our current moods played a significant impact in the kind of music we liked, and from there, we brainstormed ways to deal with this concept/problem.

What it does

Our project is a web app that allows users to input their mood on a sliding scale, and get a list of 10 curated songs that best match their mood.

How we built it

We found a database of songs that included a string of the lyrics on Kaggle. We then applied a machine learning model based on the natural language toolkit to the dataset. This formed the trained model

Challenges we ran into

As we are all beginners with full stack development, we ran into numerous errors while constructing the backend of our webpage. Many of our errors were not descriptive and it was difficult to figure out if the errors were coming from the front end, the backend or the database.

Accomplishments that we're proud of

We are most proud of getting over the challenges we faced given the strained circumstances of our work. Many of the challenges were entirely new to us and so interpreting, finding and solving these errors was a difficult and stressful process. We are very proud to have a MVP to submit.

What we learned

Working collaboratively in a high stress environment is something we are not super experienced with and it was an important lesson to learn. Given our limited full stack experience, we also learned a tremendous amount about the backend web development and using technologies like react.

What's next for Fortress

There are numerous additions we hope to make to improve the quality and functionalities of our project. Some of these include using tempo and key data to provide a stronger analysis of songs. Getting more songs in our database will help improve the quality of outputs. In addition, it would be helpful for the user to embed snippets of each song so users can listen to a small portion. Finally, it convenient feature would be exporting the song list as a Spotify playlist.