InfiniteTalk is an open-source framework for audio-driven video dubbing that supports both image-to-video and video-to-video generation. It synchronizes lip movements with audio while maintaining consistent identity, body posture, and facial expressions, even for unlimited video durations.