Skip to content

Ultravox API

Build real-time, multimodal AI applications.

Intro

The Ultravox API enables developers to build speech-to-speech AI applications. Built on the open source Ultravox LLM, the Ultravox API provides a simple REST API for creating calls with AI agents.

Ultravox REST API

We expect most applications to have a backend that will use the Ultravox REST API to create calls (and secure the API key).

The Ultravox API base url is http://api.ultravox.ai.

The API requires an API key and currently provides two endpoints:

  • Calls (/api/calls)
  • Voices (/api/voices)

Authorization

Pass in an API key using the X-API-Key header.

OpenAPI Specification

There is an OpenAPI spec file available in the repo. This spec is also used to generate the API Reference. Additionally, you can use the Ultravox API Playground to quickly explore and test the API.

Client SDKs

While there is a RESTful API for Ultravox that is used to create calls, you will need to use one of our client SDKs to join and end calls.

There are currently three SDK flavors available:

Flutter

Flutter logoflutter add ultravox_client
Get it on pub.dev.

JavaScript

JavaScript logo npm install ultravox-client
Available in the npm registry.

Python

Python logo pip install ultravox-client
More info on PyPi

Please let us know if you need a client for a specific language or framework.

Next steps

  1. Get an API KeyApply for an API Key.

  2. Do the QuickstartBuild your first voice-enabled AI agent in minutes.

  3. Read the Docs → Learn more in the API Reference.