Get your API key

Everything outside the platform UI authenticates with an API key. This page gets you from nothing to a working request.

1. Create a key

Open API keys and choose New key.

The key is shown once, at creation. We store only a hash, so a lost key cannot be recovered — create a replacement and revoke the old one.

Keys look like sk-eesi-… and carry full access to your organization. Treat one like a password: never commit it, never put it in frontend code, and give each deployment its own so you can revoke one without taking down the rest.

export EESI_API_KEY="sk-eesi-..."

2. Find out what your environment serves

Speech tiers are opt-in per environment: a backend with no configured URL is hidden entirely. So ask the API what exists rather than hardcoding a model id that may not be enabled where you are running.

curl https://api.eesi.ai/v1/audio/models \
  -H "Authorization: Bearer $EESI_API_KEY"

{
  "data": [
    { "id": "nur-tts-v1", "kind": "tts", "tier": "paid", "supports_cloning": true },
    { "id": "nur-stt-v1", "kind": "stt", "tier": "paid", "supports_diarization": false }
  ]
}

3. Make a request

Pick a tts id from that list and synthesize something.

curl https://api.eesi.ai/v1/audio/speech \
  -H "Authorization: Bearer $EESI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nur-tts-v1",
    "voice": "Nur",
    "input": "A superintelligence built to converse, not to answer."
  }' \
  --output speech.mp3

from openai import OpenAI

client = OpenAI(api_key="sk-eesi-...", base_url="https://api.eesi.ai/v1")

with client.audio.speech.with_streaming_response.create(
    model="nur-tts-v1",
    voice="Nur",
    input="A superintelligence built to converse, not to answer.",
) as response:
    response.stream_to_file("speech.mp3")

import OpenAI from "openai";
import { writeFile } from "node:fs/promises";

const client = new OpenAI({
  apiKey: process.env.EESI_API_KEY,
  baseURL: "https://api.eesi.ai/v1",
});

const speech = await client.audio.speech.create({
  model: "nur-tts-v1",
  voice: "Nur",
  input: "A superintelligence built to converse, not to answer.",
});

await writeFile("speech.mp3", Buffer.from(await speech.arrayBuffer()));

The speech endpoints speak the OpenAI wire protocol, so the official OpenAI SDKs work unchanged — you only change the base URL. See SDKs for the EESI-specific client that also covers voices, care and usage.

What a failure means

Status	Meaning	Fix
`401`	Key missing, malformed, or revoked	Check the `Authorization` header spells `Bearer sk-eesi-…`
`404` on a model id	That tier is not enabled in this environment	Read `/v1/audio/models` and use an id from it
`402`	Credits exhausted	Top up in Billing
`429`	Rate or concurrency limit for your plan	Retry with backoff, or raise the plan

Text to speech

Formats, streaming, speed, and voice selection.

Voices

Clone a voice, and why every voice is a reference clip.

Realtime

Speech-to-speech over a WebSocket.

Authentication

Scopes, rotation, and how keys differ from session tokens.

Get started

Speech

Realtime

Telephony

Care

Platform

Get your API key

1. Create a key

2. Find out what your environment serves

3. Make a request

What a failure means

Next

Text to speech

Voices

Realtime

Authentication

​1. Create a key

​2. Find out what your environment serves

​3. Make a request

​What a failure means

​Next

Text to speech

Voices

Realtime

Authentication

1. Create a key

2. Find out what your environment serves

3. Make a request

What a failure means

Next