Gemini 2.0 Flash Enterprise & Real-Time Multimodal Live API
Google’s Gemini 2.0 Flash introduces native real-time bidirectional streaming over WebSockets, allowing applications to process continuous microphone audio, camera video, and text output with under 300ms latency.
---
🌐 Real-Time Audio/Video WebSocket Architecture
[Browser / Mobile App] -- (PCM 16kHz Audio / WebM Video Frames) --> [WebSocket Connection]
|
[Gemini 2.0 Live API Engine]
|
[UI Audio Player] <--- (24kHz Audio Chunk PCM + Function Call) ---------+
🛠️ Code Pattern: Initializing Google GenAI Live Client
import { GoogleGenAI } from '@google/genai';
const ai = new GoogleGenAI({ apiKey: process.env.GEMINI_API_KEY });
// Connect to Gemini Live API
const session = await ai.models.connectLiveSession({
model: 'gemini-3.1-flash-live-preview',
config: {
generationConfig: {
responseModalities: ['AUDIO', 'TEXT'],
speechConfig: {
voiceConfig: { prebuiltVoiceConfig: { voiceName: 'Puck' } }
}
}
}
});