DeepTunes was the first cross-functional build week project at Lambda School. We had only four days to work with a brand new team of web developers and data scientists and create a fully functional web app.
We were one of two teams selected out of 15+ to present at “Shark Tank,” a school-wide presentation of 300+ people.
Check out the Shark Tank demo here: https://youtu.be/5P3oDO8v0Sc
Have you ever listened to a track on Spotify, then had your Discover Weekly populate with similar tracks? That’s what it’s designed to do. But sometimes your partner listens to a track on your account, or you go down a rabbit hole of a single artist- and your Discover Weekly reflects this, perhaps not in a way you want.
We created DeepTunes to take the personalization out of recommendations- an incognito tab for your Spotify. We took a database of hundreds of thousands of songs and created a recommendation engine that gives you similar songs based on the song’s attributes, not on your listening history.
A user inputs a single song that they like, and we generate a list of similar songs based on the features of our dataset: Duration, Key, Mode, Time Signature, Danceability, Instrumentalness, Liveness, Speechiness, and Valence.
We also visualize these attributes, something that Spotify doesn’t allow you to do.
Street Smarts was an incredible week-long project with a great team. We learned a ton about recommender systems and working cross-functionally.
Originally published at https://willstauffernorris.github.io.