Skip to content

Smart Glasses Architecture

The Omni Smart Glasses client runs on ESP32-based wearable devices, providing hands-free voice interaction.

Hardware prototype

This component is a hardware prototype. See the smart-glasses/ directory for schematics and firmware.

Overview

The smart glasses provide:

  • Voice input — Onboard microphone captures speech
  • Audio output — Bone conduction speaker for private audio
  • WiFi connectivity — Connects to the Omni backend over WebSocket
  • Minimal display — Optional small OLED/LED for status indication

Communication Flow

sequenceDiagram
    participant G as Smart Glasses
    participant BE as Backend
    participant GEMINI as Gemini Live API

    G->>BE: WebSocket + PCM audio
    BE->>GEMINI: Bidi streaming
    GEMINI->>BE: Audio response
    BE->>G: PCM audio playback