Speech

Create speech (OpenAI-compatible)

Generate audio from text in OpenAI's text-to-speech format

POST/v1/audio/speech

Generate audio from text using the format of OpenAI's /v1/audio/speech endpoint. To use the OpenAI SDKs with Inworld, set the base URL, API key, model, and voice to Inworld values. See OpenAI Compatibility for SDK examples, audio formats, and streaming.

Authorizations

Authorizationstringrequired

Your authentication credentials. Pass your API key as a Bearer token: Bearer $INWORLD_API_KEY. The OpenAI SDKs send the key this way.

Body

application/json

modelstringrequired

An Inworld model ID. See Models. OpenAI model names such as tts-1 are rejected.

inputstringrequired

The text to synthesize, up to 4,000 characters. Inline markup such as steering tags and pause controls is supported.

voicestringrequired

An Inworld voice ID, or the ID of one of your custom voices. Voice IDs are case-sensitive. OpenAI voice names such as alloy are not available.

response_formatenum<string>default: "mp3"

The audio format. mp3, opus, flac, and wav are 48 kHz mono. pcm is raw 16-bit signed little-endian mono samples at 24 kHz, with no header. aac is not supported.

Available options:mp3opusflacwavpcm

speednumberdefault: 1

The speaking rate. Values outside 0.5 to 1.5 are rejected.

instructionsstring

A natural-language steering instruction for the whole request. Supported on inworld-tts-2 only; other models ignore it.

stream_formatenum<string>default: "audio"

audio streams raw audio bytes. sse streams server-sent events.

Available options:audiosse

Create speech

curl -X POST https://api.inworld.ai/v1/audio/speech \
  -H "Authorization: Basic $INWORLD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "inworld-tts-2",
    "voice": "Dennis",
    "input": "Welcome back! Your order shipped this morning and should arrive by Thursday."
  }' \
  --output speech.mp3

default_response

"string"