EESI builds voice models for real conversations. Nur listens and answers as you talk, Live translation interprets as people speak, and the speech APIs turn text into audio and audio into text. Audio intelligence reads a recording the way an editor would: who spoke, how each line landed, and what happened around the voice. Everything runs on one API with one key, at https://api.eesi.ai/v1. Most of it speaks the OpenAI wire format, so existing clients work with a new base URL.

Quickstart

A key, a voice, and your first conversation in a few minutes.

Nur

A spoken conversation over WebRTC or a WebSocket, with interruptions handled for you.

Live translation

Speak one language and be heard in another, in your own voice if you like.

Text to speech

Speech in a built-in voice or one cloned from a short clip.

Speech to text

Transcripts with speakers and word timings.

Audio intelligence

Delivery, reactions, key moments and stems for any recording.

​
Start from what you’re building

Mobile apps

Talk to Nur from your app over WebRTC, without shipping your key.

Contact centers

Nur answers calls behind your own phone system.

Phone calls

Nur calls someone and holds the conversation you describe.

Game characters

Characters players talk to, with live game state and tools.

Robots

A robot that talks, gets cut off gracefully and acts while it speaks.

Your website

A voice companion on your site with one script tag.

​
The model

One model, nur-v1, does all of it. Send it to the endpoint for what you need: Nur speaks and understands more than a hundred languages, and switches when the person it is talking to does.

​
Try it first

Talk to Nur in 100+ languages, including 19 Arabic dialects, at languages.eesi.ai. No account needed. Sign in to the console to try Live, translation and the speech tools in your browser and to create API keys.