Introduction

Voice integration lets you embed an Elizon voice agent into your own application — web, mobile, or native — so your users can talk to it directly, without going through the Elizon Studio UI.

What voice integration is

An Elizon agent configured with a voice configuration (TTS + STT providers) can hold a live, two-way audio conversation over LiveKit, an open-source WebRTC platform. Your application discovers which of your agents support voice, starts a session for one of them, and connects to the resulting LiveKit room using a short-lived participant token. From that point on, the agent listens, responds, and speaks back in real time over the same connection.

The two-step flow

Every voice integration follows the same two calls against the Elizon Public API, then a client-side connect:

1. Discover — GET /api/v1/voice-agents lists the agents in your organization (or project) that have a voice configuration, so you know which agentId to use.
2. Start a session — POST /api/v1/voice-agents/:agentId/sessions creates a LiveKit room, dispatches the voice agent into it, and returns a wsUrl + participant token.
3. Connect — pass that wsUrl and token to a LiveKit client SDK (room.connect(wsUrl, token)) to join the room and start talking.

Everything after step 3 — audio capture, playback, mute/unmute, disconnect — is handled by the LiveKit client SDK for your platform.

Where to go next

• Quickstart — discover an agent and start your first voice session with curl.
• SDK Integration — connect from Web, React, iOS, Android, React Native, Flutter, or Unity.