AI/ML / COMPLETED
ViNi: A voice and camera assistant with Gemini Live
A browser-based robot companion built with the Gemini Live API, featuring real-time voice, camera tools, touch-to-wake interaction, and selectable personalities.
THE IDEA
Why I built it
I built ViNi to explore a voice-first AI companion with a robot face, touch interaction, and camera tools.
Project goals
- Combine live audio with camera input
- Create a mobile-friendly progressive web app
- Keep API keys on the server
Challenges
- Managing latency and echo cancellation
- Proxying WebSocket connections safely
What came out of it
- A ViNi demo and public source code