OptionalappliesWhether the service honours the OpenAI-style speed field (default true). Set false for
one that ignores it: the editor then sends speed 1 and plays the clip at the chosen speed,
keeping the voice's pitch.
An OpenAI-compatible /audio/speech URL — your own server or proxy (it adds the key), a
hosted service, or a self-hosted open voice model with that API.
OptionalfetchFor tests / custom transports.
Optionalinit: RequestInitOptionalinit: RequestInitOptionalformatAudio format asked for (default mp3).
OptionalheadersOptionalmodelModel name sent as model.
OptionalvoicesThe voices to offer. Default: the six common OpenAI-style voice ids.
Options of createSpeechEndpoint.