If there's one thing I think Spotify can improve on, it's Spotify radio. It often gives recommendations that are songs I already have saved, or just seem completely unrelated to the original song! So I decided to do something about it.
Using the Spotify API, I utilized song attributes such as its lyrics, tempo, liveness, energy, etc..., and provided more accurate recommendations based on the inputted song. Just for fun, I made it so I'm able to track what my top artists/songs were in the last month, six months, and year.
This was inspired after I was trying to find more songs similar to "love." by wave to earth. But again, Spotify radio kept giving me tracks that I already knew or just weren't related at all. This app I developed allowed me to find tracks that actually sounded just like it and weren't just some random track recommendation based on my overall listening activity.
Of course, recommendations aren't always perfect, so I made it so that I could refresh the results to give me a set of new tracks.
This fullstack app was built with React on the frontend, and an Express.js server powered by Node.js on the backend. It utilizes a proper OAuth 2.0 flow with both implicit grants and/or token based logins to access user data and play tracks on the webpage.