Docs
Speech synthesis
Turn text into speech, for guidance prompts and reading out notifications.
Calling
POST /audio/speech with input and voice. The response is an audio binary.
| Field | Type | Note |
|---|---|---|
| input | string | Text to speak |
| voice | string | Model dependent |
| response_format | string | mp3 | wav | pcm |
| speed | number | 0.5–2.0 |
Streaming
Set stream_format to send audio as it is produced, reducing perceived latency on long passages.
Delivery control
Adjust pace with speed. Punctuation maps directly onto pauses, so it is better to edit the source text.