TTS overview

Turn text into natural, expressive speech with Inworld TTS, and make your first request.

Inworld TTS turns text into natural, expressive speech. Give your app, agent, or character a voice, narrate long content, or clone and design voices. The Realtime TTS-2 models speak 200+ languages, stream the first audio in tens of milliseconds, and let you steer emotion and delivery in natural language.

Quickstart

Follow the steps to make your first API call.

TTS Playground

Try voices, models, and instant voice clones in the Portal.

Code examples

Find code examples for common use cases.

Using AI to code? Give your assistant the docs index at https://docs.inworld.ai/llms.txt. For live search, add the MCP server.

Prefer the terminal? Install the Inworld CLI with npm install -g @inworld/cli. Use it to generate speech and audiobooks, or clone and design voices. AI agents can use it too.

Your first request

Install the SDK with npm install @inworld/tts or pip install inworld-tts. Set INWORLD_API_KEY to the Base64 credential from the Portal, then generate an audio file:

JavaScript

import { InworldTTS } from '@inworld/tts';
import fs from 'fs';

const tts = InworldTTS();

const audio = await tts.generate({
    text: 'What a wonderful day to be a text-to-speech model!',
    voice: 'Ashley',
});

fs.writeFileSync('output.mp3', audio);

Python

from inworld_tts import InworldTTS

tts = InworldTTS()

tts.generate(
    text="What a wonderful day to be a text-to-speech model!",
    voice="Ashley",
    output_file="output.mp3",
)

The quickstart explains this request step by step. It also covers streaming, which we recommend for realtime applications.

Explore

Playground guide

Learn how to use the Portal playground.

Models

Realtime TTS-2 and TTS-2 Flash, with a feature comparison.

Generating audio

Voices, languages, formats, and context for natural delivery.

Instant voice cloning

Clone a voice from a short sample.

Steering

Direct emotion, pace, and style in natural language.

Synthesize speech

Choose a synchronous, streaming, WebSocket, async, or batch endpoint.

Best practices

Make speech expressive, fast, and ready for production.

To generate text with an LLM and speak it in one API call, use Voice responses. For a live two-way conversation, use the Realtime API.