SoundMatch initially began as a final project for our Computer & Society class.
Like most students, we use music streaming services daily, which sparked our curiosity about how platforms like Spotify's "Discover Weekly" and Apple Music know exactly what we want to hear next.
As Computer Science majors, we wanted to apply the programming and math concepts we learned in classes like Discrete Structures to understand and implement content-based filtering for ourselves.
Our first major hurdle was finding a good, clean dataset with song features, especially since raw text data often contains a lot of noise.
From there, the main challenge was defining what makes two songs "similar" using only data, which meant figuring out how to analyze songs as vectors of audio features to actually capture a song's vibe.
Overcoming these technical hurdles and seeing the math work in action turned this from just a school assignment into a personal project we were genuinely interested in.
We eventually decided to expand it beyond our original custom GUI into a full, public-facing website so that others could easily use it to discover new music based on their own unique tastes.