Released in mid-2025, this closed-source, highly specialized V2V model's input consists of an existing video clip and a new audio track. The model analyzes the audio and maps phonemes to the facial region of the video, generating new mouth movements that are blended into the original footage. It prioritizes precise lip-sync accuracy over any other editing, leaving the rest of the video untouched. This makes it ideal for dubbing dialogue or matching pre-existing footage to different audio. Consider alternatives like Sync Labs for similar workflows.