Gemini Live Interviewer Demo

A real-time interviewer experience using Google's Gemini Live API with the official @google/genai SDK. Talk to Gemini using your microphone and hear responses in real-time with text transcription.

Features

• Real-time bidirectional voice communication with Gemini
• Text transcription of Gemini's responses
• Text message input option
• Browser-based microphone capture with device selection
• Audio streaming with WebSockets
• Modern, clean UI with mute and hangup controls
• Google Search integration

Prerequisites

• Node.js (v18 or higher recommended)
• A Google API key from Google AI Studio

Setup

1. Install dependencies:

   npm install

2. Configure API key:

   • Copy .env.example to .env
   • Add your Google API key to the .env file:

     GOOGLE_API_KEY=your_actual_api_key_here
     

3. Start the development server:

   npm run dev

4. Open in browser:

   • Navigate to http://localhost:8000
   • Select your microphone from the dropdown (bottom right)
   • Click the microphone button to start
   • Allow microphone access when prompted
   • Start talking to Gemini!

How It Works

1. Frontend (index.html):

   • Captures microphone audio at 16kHz, 16-bit PCM, mono using AudioWorklet
   • Converts audio to base64 and sends via WebSocket to backend
   • Receives and plays audio responses from Gemini at 24kHz
   • Displays text transcription in real-time
   • Provides text input for messaging
   • Mute/unmute and hangup controls

2. Backend (server.ts):

   • Built with Hono framework for HTTP server
   • Manages WebSocket connections from clients
   • Connects to Gemini Live API using @google/genai SDK
   • Forwards audio data between client and Gemini
   • Handles response streaming and transcription
   • Broadcasts audio and text to all connected clients

Technical Details

• Audio Format (Input): 16-bit PCM, 16kHz, mono
• Audio Format (Output): 24kHz from Gemini
• Model: gemini-live-2.5-flash-preview
• Voice: Zephyr
• Response Modalities: AUDIO with output transcription
• Tools: Google Search enabled
• Framework: Hono + Node.js WebSocket Server
• Runtime: tsx for TypeScript execution

Features Explained

Voice Chat

• Click the microphone button to start a live conversation
• Use mute button to temporarily disable your microphone
• Use hangup button to end the session and clear audio

Text Input

• Type messages in the text field and press Enter
• Gemini will respond with audio and text

Microphone Selection

• Use the dropdown in the bottom-right corner to select your audio input device

Troubleshooting

• No audio received: Check your API key is valid and has access to the Gemini Live API
• Microphone not working: Ensure browser has microphone permissions and a device is selected
• Connection errors: Verify your API key in the .env file
• Module errors: Ensure you have Node.js v18+ and run npm install again
• Port already in use: The server runs on port 8000 by default, ensure no other service is using it

Credits

Based on the work by jaydanurwin

API Documentation

For more information about the Gemini Live API, see:

• https://ai.google.dev/gemini-api/docs/live
• https://github.com/google-gemini/multimodal-live-api-web-console