In a variety of human-machine interaction (HMI) applications, the high-level techniques based on audio-driven talking face generation are often challenged by the issues of temporal misalignment and ...