Abstract: Starting from the strategic needs of the national new industrialization, combined with the development trend of “immersion” in the meta-universe, this study makes an in-depth analysis of the immersive experience of digital intelligence performance and its characteristics in different evolution forms, and brings up a comprehensive design method which integrates immersive cross-scene experience with multimodal collaborative interaction. Specifically, it is suggested to construct a collaborative human-computer interaction between the audience and the virtual avatar, and adopt a universal cross-scene narrative interaction design method. Through the collaboration and decision-making of the multimodal human-computer interface, the close connection of emotional perception between the audiences can be realized. In addition, the study designs and develops a performance prototype system for practice and specific elaboration, expands the natural interaction in the immersive experience of digital intelligence performance, and improves the effect of audience experience with the actual experience of the audience. This opens up a new way for the presentation form of meta-universe performance in the intelligent era and provides feasible suggestions for implementation.