Skip to main content
Omnia Voice is a platform for building voice AI agents — AI that holds a real spoken conversation over the phone or in the browser, calls your APIs mid-call, and answers from your own documents. You configure an agent once — personality, voice, language, tools, knowledge — and then reach it however you need: an inbound phone number, an outbound call, a widget on your website, or a raw WebSocket from your own application.

Quickstart

Four paths — browser, inbound phone, outbound, or your own telephony.

API reference

Every endpoint, with an interactive playground.

Tools

Let an agent call your APIs and control the call.

Worked examples

Complete builds, start to finish.

​
What the platform gives you

Agents

Personality, voice, language, call behaviour.

Tools

HTTP, client, and built-in call control.

Corpora

Answer from your own documents.

Telephony

Our numbers, or bring your own carrier.

SDKs

JavaScript and React, or raw WebSocket.

Transcription

Batch or streaming speech to text.

​
What you can build

  • Inbound reception — answer every call, day or night, in the caller’s language
  • Outbound campaigns — reminders, follow-ups, qualification calls
  • In-app voice — a spoken assistant inside your web product
  • Transcription pipelines — batch or real-time speech to text, no agent involved

​
Two products, one API

Both share the same API key and the same account. Use either on its own, or both together.

​
How a call works

1

You create an agent

An agent bundles a prompt, a voice, a language, call settings, and any tools or knowledge bases it should have access to.
2

Someone reaches it

Through a phone number you have assigned, an outbound call you trigger, the embeddable web widget, or a WebSocket session you open yourself.
3

The agent handles the conversation

It listens, decides what to say, and calls your tools when it needs real data — checking an order, booking a slot, looking something up.
4

You get the record

Transcript, recording, duration, and cost are all available through the API and in the dashboard once the call ends.

​
Base URL

https://dashboard.omnia-voice.com/api/v1 points at the same service and keeps working. New integrations should prefer api.omnia-voice.com.

​
Your first request

​
Languages and voices

Agents support 50+ languages, with particular depth in English and the Nordic languages. Language is auto-detected during a call, and an agent can switch mid-conversation without being told to. Fetch what is available for your account with GET /languages and GET /voices — you will need a languageId and a voiceId to create an agent.