Built at the UC Berkeley AI Hackathon 2025 | June 21–22, 2025
🔗 Live App: crescendoai.vercel.app
📄 Devpost: devpost.com/software/crescendo-yt3wc1
📂 Original Repo: github.com/Victoria-DR/uc-berkeley-ai-hackathon-2025
Crescendo AI is an intelligent music performance analysis tool that helps musicians improve by providing real-time, AI-driven feedback on their playing. Whether you're practicing piano, guitar, or any other instrument — Crescendo AI listens, analyzes, and tells you exactly where you went right and where you can improve.
- 🎤 Live Listening — Play your instrument live and Crescendo AI captures your performance in real-time
- 📁 Upload a Recording — Already have a recording? Upload it directly for analysis
- 🤖 AI Analysis — Google Gemini processes your performance and evaluates technique, timing, note accuracy, and more
- 📊 Detailed Feedback — Get a breakdown of what you played well and specific areas where mistakes were made, with suggestions for improvement
| Layer | Technology |
|---|---|
| Frontend | Next.js, React, JavaScript, CSS |
| AI/ML | Google Gemini API |
| Audio Processing | Web Audio API |
| Deployment | Vercel |
Built in 24 hours by a team of 4 at UC Berkeley:
| Name | Role |
|---|---|
| Aakash | AI/ML Integration & Full-Stack Development |
| Victoria | Frontend Development |
| Janet Meng | Development |
| Sujal | Development |
As the AI/ML integration lead and a full-stack contributor, I was responsible for:
- Gemini API Integration — Architected and implemented the connection between the audio input pipeline and Google Gemini for intelligent music analysis
- Audio Processing Pipeline — Worked on capturing and processing live audio and uploaded recordings for AI consumption
- Performance Feedback Engine — Developed the logic that translates Gemini's analysis into actionable, musician-friendly feedback
- Full-Stack Development — Contributed across the entire stack, from UI components to backend API routes
- Node.js 18+
- A Google Gemini API key
# Clone this repository
git clone https://github.com/YOUR_USERNAME/crescendo-ai.git
cd crescendo-ai
# Install dependencies
npm install
# Set up environment variables
cp .env.example .env.local
# Add your GEMINI_API_KEY in .env.local
# Run the development server
npm run devOpen http://localhost:3000 to see the app.
- Event: UC Berkeley AI Hackathon 2025 (3rd Edition)
- Organized by: Cal Hacks
- Sponsors: Anthropic, Google, Fetch AI, and more
- Duration: 24 hours (June 21–22, 2025)
- Location: Martin Luther King Jr. Building, UC Berkeley
This project was built during a hackathon and is open for educational and portfolio purposes.
Made with ☕ and 🎵 at UC Berkeley AI Hackathon 2025