Text to Speech with MessagePack

For better speech quality and lower latency, upload reference audio via the create model endpoint. This method doesn’t require MessagePack and uses this API instead.