Docs/Official SDKs

Official SDKs.

Maintained by us, kept feature-flat with the REST and WebSocket APIs, every release pushed within a week of the underlying change.

Python

Synchronous and async clients. Streaming via aiohttp; works inside FastAPI, Django Channels, and plain scripts.

bash

pip install celestlabs

python

from celestlabs import CelestLabs

client = CelestLabs()  # reads CELESTLABS_API_KEY from env

# one-shot
audio = client.speech.create(voice="meera", input="Hello, world.")

# streaming
with client.realtime.connect(voice="arjun") as session:
    session.send_text("Streaming voice in real time.")
    for chunk in session.audio_chunks():
        speaker.write(chunk)

Node.js / TypeScript

bash

npm i @celestlabs/sdk

javascript

import { CelestLabs } from "@celestlabs/sdk";

const client = new CelestLabs();  // reads CELESTLABS_API_KEY

const audio = await client.speech.create({
  voice: "meera",
  input: "Hello, world.",
});

// streaming with an async iterator
for await (const chunk of client.realtime.stream({ voice: "arjun", input: "Real time." })) {
  speaker.write(chunk);
}

Go

bash

go get github.com/celestlabs/celestlabs-go

Ruby

bash

gem install celestlabs

cURL recipes

Useful for shell scripts, CI, and quick sanity checks.

bash

# Save synthesised WAV
curl -sS https://api.celestlabs.ai/v1/audio/speech \
  -H "Authorization: Bearer $CELESTLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"voice":"meera","input":"Hello."}' \
  --output hello.wav

# Pipe directly into ffplay for instant playback
curl -sS https://api.celestlabs.ai/v1/audio/speech \
  -H "Authorization: Bearer $CELESTLABS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"voice":"saanvi","input":"Hello.","format":"mp3"}' \
  | ffplay -nodisp -autoexit -

OpenAI SDK works too

If you'd rather not add another dependency, the openai SDK with base_url swapped to ours covers the REST surface. See OpenAI compatibility.

← Previous

Rate limits

Changelog