So far sdialog.audio only supports Kokoro or Hugging Face TTS. It would be good to allow the user to use API-based TTS