Inference

Synthesize speech from text via ElevenLabs (mp3)

Returns audio bytes (default audio/mpeg) generated by ElevenLabs. Billed per character with a fixed 30% margin on the provider base cost — irrespective of the user's subscription tier. Usage and cost are returned via X-OpenHuman-* response headers.

post/openai/v1/audio/speech

Request

  • Base URL: https://api.tinyhumans.ai
  • URL: https://api.tinyhumans.ai/openai/v1/audio/speech
  • Auth: HTTP bearer

Request body

textstring required
voice_idstring

Voice id. Defaults to ELEVENLABS_DEFAULT_VOICE_ID.

model_id'tts-v1'

Stable, provider-agnostic TTS model id. Mapped to the underlying upstream model internally.

output_formatstring

ElevenLabs output format (e.g. mp3_44100_128).

with_visemesboolean

When true, returns JSON containing base64 audio plus a viseme timeline (Oculus 15-viseme set) and the raw character-level alignment from ElevenLabs. When false (or unset), the response is raw audio bytes.

Example request

{
  "text": "The first move is what sets everything in motion."
}

Response

Audio bytes

Changes

No recorded changes to this operation across all 1 revision of this API.