Mason Scott
🎵 Impulse – Swipe-Based Music Discovery Using Spotify
A full-stack music recommendation app that lets users swipe through personalized song suggestions and sync their likes directly to Spotify. Built with React, Spring Boot, and a FastAPI-powered recommendation engine.
🚀 Demo
🎥 
📌 Note: This app is still in development. Some features are using hardcoded demo data to simulate final behavior.
🔧 Development Progress
Frontend (React) [█████████████———] 85%
Backend (Spring Boot) [██████████————] 75%
Recommendation Engine [█████████████████████] 99%
🧠 Core Features
- 🎵 Swipe-based interface for music discovery (like/dislike)
- 🔐 Spotify login required – users must have a Spotify Premium account
- 📈 User stats tracking, dark mode toggle, and interactive settings menu
- ❤️ Liking a song in the app automatically adds it to your Spotify liked songs
- 📦 Fully Dockerized for seamless deployment on a VPS
🖥️ Tech Stack
| Layer | Technology |
|---|---|
| Frontend | React, SCSS |
| Backend API | Spring Boot (Java) |
| Recommendation | FastAPI (Python), Celery, Redis, MySQL |
| Deployment | Docker Compose |
| Auth / Playback | Spotify Web Playback SDK, OAuth 2.0 |
🧠 About the Recommendation Engine
The recommendation engine powering Impulse is built using FastAPI and lives in a separate sub-repo:
🔗 Recommendation API Repo
It features:
- Audio embedding via
librosa - Semantic similarity search via cosine distance
- Background processing with Celery & Redis
- YouTube fallback for audio preview links
For full details on how the engine works, see the Recommendation API README
📌 Notes
- All songs shown in the demo are hardcoded to simulate the app flow.
- In the final version, all playback and song data will be dynamically pulled using the Spotify SDK and personalized based on user listening history.
- The system is fully Dockerized and built for modular development across services.
👤 Author
Mason Scott
Third year CS major (Data Science track), Statistics minor
University of Maryland – College Park
🌐 masonscott.net
🐙 GitHub
🔗 LinkedIn
📄 License & Attribution
This app uses the Spotify Web API and the Spotify Web Playback SDK.
Impulse is not affiliated with or endorsed by Spotify.