Zyra
Interactive art player that uses AI only to fetch YouTube URLs, runs them through yt-dlp, and turns playback into touch-responsive visuals.
Capstone Project · Capstone Project (2025) · Developer
Zyra uses a lightweight AI agent strictly to parse natural language prompts and return the precise YouTube URL a user is looking for.
Once the link is confirmed, yt-dlp handles downloading and converting the audio so it can be played locally with consistent latency.
Art Mode then launches a full-screen canvas with shaders, particle fields, and typography that react to both the beat and direct touch or pointer gestures.
The goal is to jump from an idea to an immersive audio-visual performance in seconds without AI trying to compose music—just to fetch the right source.
Features
- AI prompt interpreter that returns the exact YouTube URL users actually want
- YouTube metadata parsing with automatic yt-dlp download + conversion pipeline
- Audio playback controls with seamless buffering and track switching
- Touch/gesture-driven Art Mode that paints interactive visual layers
- Audio-reactive visualizer that syncs color, motion, and typography to the beat
- Visualizer presets for concerts, neon grids, and experimental typography modes
Built with React, Next.js, yt-dlp, TypeScript, WebGL, Canvas, Audio API.
