Custom text-to-speech configurations aren’t applied when saving the agent. Instead, general text-to-speech configurations in Settings are used. If I want a particular voice or setting for an agent, I need to repeatedly change the general settings for each agent.