750 ms is really good! I've heard multilingual assistants have higher latency. But if your assistants are responding quickly your next thing to improve is the script