Undertone

💡 Inspiration

We’ve all experienced the "algorithmic echo chamber": you listen to one Jazz track, and your streaming service traps you in a loop of the same three sub-genres forever. Most recommendation systems rely on Collaborative Filtering, which suggests music based on what other people liked, rather than how you actually feel.

Undertone was born to break the loop. We believe technology should enhance the human experience by resonating with the heart. We wanted to build an Intent Engine that treats music as art—prioritizing the objective physical properties of sound and the subjective context of emotion over mere tags and history.

🚀 What it does

Undertone is a high-performance music discovery utility that separates "Mainstream" noise from "Niche" soul to find your true next favorite track.

1. Intent-Based Search: Instead of searching by genre, you search by feeling.
2. Acoustic Precision: It parses objective physical properties—BPM, Loudness, and Frequency—to match your current energy level.
3. Echo-Chamber Breaker: A dedicated discovery logic that strictly isolates unknown artists from the "Mainstream" bubble.
4. Subjective Tagging: A community-driven context engine that understands that "sadness" in a lo-fi track feels different than "sadness" in a blues record.

🛠️ How we built it

We prioritized Algorithmic Integrity and a Minimalist Aesthetic. The stack was chosen for precision and speed:

• Intelligence: A Decoupled Monolith architecture using Flask for rapid algorithmic iteration and SQLAlchemy ORM for complex data relationships.
• Signal Processing: We integrated Librosa for real-time validation of musical attributes, allowing the system to "understand" bass intensity and screaming logic programmatically.
• Frontend: Built with a high-performance Navy Blue design system using Vanilla JS and CSS Variables for a premium, low-latency user experience.
• Data Integrity: Dynamic Filter Chips construct real-time queries for complex intent matching, ensuring the interface stays reactive and humane.

🧠 Challenges we ran into

The most significant hurdle was the "Subjectivity of Sound." Defining "intensity" is easy for a computer; defining "energy" or "vibe" is not. We had to build a validation layer that could distinguish between a high-BPM track that is "calm" (like liquid DnB) and one that is "aggressive" (like Hardcore Punk).

Another major challenge was Data Density. Managing structured relationships between thousands of songs, user libraries, and subjective tags required intense database optimization to keep the Intent Engine feeling instantaneous.

💪 Accomplishments that we're proud of

We are incredibly proud of our Signal Validation Engine. Seeing the app successfully differentiate between "Mainstream" patterns and truly "Niche" acoustic properties was our "Eureka" moment. We built a system that doesn't just look at what you’ve heard—it reasons about what you’ll love next.

📚 What we learned

This project reinforced that the future of personal software isn't about more data—it’s about Synthesis. We learned how to manage complex Signal Processing flows while keeping the UI approachable. Technology is a tool, but subjectivity—art and music—is what truly changes the world.

🗺️ What's next for Undertone

• Real-time Spotify/Apple Music Sync: To move from discovery to immediate listening.
• Collaborative Vibes: Private rooms where groups can generate a "collective intent" playlist.
• Waveform Visualization: Bringing the "Rich Aesthetic" further with real-time audio visualizers in the UI.

Crafted on the bleeding edge of the Python & Flask Ecosystem.

Built With

• css
• html
• javascript
• python