Sync a face's mouth to an audio track across four lip-sync endpoints.

• Routes intelligently between Sync Labs (mouth-swap on existing video), OmniHuman (portrait + audio to talking-head avatar), Kling lipsync (audio-to-video or text-to-video with in-pass speech generation), and Creatify, picking the right model based on input type and quality tier.
• Handles three workflows: source video + audio for mouth replacement, portrait still + audio for avatar-style generation, and script-only for in-pass audio synthesis.
• Invokes via runcomfy run CLI with JSON input; automatically detects user intent from natural language triggers like "lip sync," "dub video," "sync lips to voice," or explicit model names.
• Requires runcomfy CLI authentication and enforces consent guardrails—refuses requests targeting real public figures without consent or aiming at defamatory/explicit synthetic media.