why do you need a server endpoint ? you should just be able to set the voice parameters like transcription service and model