MultiTalk is an open-source framework for generating multi-person conversational videos from audio, reference images, and prompts. It supports expressive animation, lip-syncing for multiple characters, and various interaction controls for stylized and realistic video outputs.