Building Real-time Voice
Applications with LiveKit

What is LiveKit?

LiveKit is an open-source, end-to-end WebRTC platform. It abstracts away the notoriously complex WebRTC stack and gives you clean APIs for building real-time audio and video applications. Think of it as the "Stripe of real-time communication" — it handles the hard infrastructure so you can focus on your product.

Why Not Just Use WebRTC Directly?

You could build on raw WebRTC. But you'd spend months dealing with STUN/TURN servers, codec negotiation, bandwidth estimation, connection state machines, and the dozens of browser-specific quirks that make WebRTC notoriously hard to ship.

LiveKit gives you production-ready infrastructure out of the box:

Voice-Specific Superpowers

LiveKit isn't just a generic WebRTC wrapper. It has first-class support for voice applications with features that would take months to build yourself.

Building a Voice Chat Room

Let's walk through the key pieces. A voice application has two sides: a backend that generates secure tokens, and a frontend that connects to the room and handles audio.

  Client App
      │
      ▼
  LiveKit SDK  ──→  Token Server (your backend)
      │
      ▼
  LiveKit SFU  ──→  Routes audio streams
      │
      ▼
  Other Clients

Step 1: Token Generation (Backend)

Every participant needs a JWT token to join a room. This is generated server-side so your API keys are never exposed to clients.

import { AccessToken } from 'livekit-server-sdk';

function createToken(roomName: string, participantName: string) {
  const token = new AccessToken(apiKey, apiSecret, {
    identity: participantName,
  });

  token.addGrant({
    room: roomName,
    roomJoin: true,
    canPublish: true,     // Can send audio
    canSubscribe: true,   // Can receive audio
  });

  return token.toJwt();
}

Step 2: Room Connection (Frontend)

On the client side, create a Room instance, configure audio settings, and connect. LiveKit handles all the WebRTC negotiation, ICE candidate gathering, and codec selection behind the scenes.

import { Room, RoomEvent } from 'livekit-client';

const room = new Room({
  adaptiveStream: true,
  dynacast: true,
  audioCaptureDefaults: {
    autoGainControl: true,
    echoCancellation: true,
    noiseSuppression: true,
  },
});

// Someone joins
room.on(RoomEvent.ParticipantConnected, (participant) => {
  console.log(`${participant.identity} joined the room`);
});

// Receive their audio
room.on(RoomEvent.TrackSubscribed, (track, pub, participant) => {
  if (track.kind === 'audio') {
    const audioElement = track.attach();
    document.body.appendChild(audioElement);
  }
});

// Connect and enable mic
await room.connect(livekitUrl, token);
await room.localParticipant.setMicrophoneEnabled(true);

Production Best Practices

Getting a demo working is one thing. Shipping to production is another. Here are the things that matter when real users are on the line.

Beyond Basic Voice Chat

Once you have the fundamentals, LiveKit opens the door to some genuinely exciting use cases.

Performance Numbers That Matter

When building voice applications, these are the benchmarks you should be targeting. Miss any of them and your users will notice.

Deployment Options

LiveKit gives you flexibility in how you deploy, each with different tradeoffs.