Voices and cloning
List voice profiles, engine presets and cloning support.
GET /v1/audio/voices connects VoiceStudio's voice library to
OpenAI-compatible clients.
curl "$VOICESTUDIO_BASE_URL/audio/voices" \
-H "Authorization: Bearer $VOICESTUDIO_API_KEY"The response has two lists:
voices: OpenAI aliases and your saved profiles.engines: installed TTS engines, with cloning support, device routing and setup hints.
List just your profiles:
curl -s "$VOICESTUDIO_BASE_URL/audio/voices" \
-H "Authorization: Bearer $VOICESTUDIO_API_KEY" \
| jq '.voices[] | select(.type == "profile") | {voice_id, name, language}'Pass a voice_id straight into speech generation:
{
"model": "tts-1",
"voice": "demo0001",
"input": "This line uses a saved VoiceStudio profile."
}What voice accepts
| Value | Result |
|---|---|
default | The selected engine's default voice |
alloy, echo, fable, nova, onyx, shimmer | Accepted for compatibility; mapped to the engine default |
| A profile ID | The saved reference audio, transcript, style and seed |
| An engine preset | Passed to engines that ship named voices |
Create, lock, import, export and manage consent for profiles with the
/profiles/* and /personas/* routes, documented in the
API reference.
Local IDs
Profile IDs here belong to the desktop app. The unreleased Cloud contract uses tenant-owned Voice resources, and a local profile ID or request body won't carry over unchanged.