In order to solve the problem of insufficient brand communication efficiency and content adaptation in new media short video platform, this paper constructed an AIGC short video generation and market application analysis framework integrating text, visual and audio information. Multi-modal features are extracted from brand short video samples, comment corpus, interaction log and product attributes. After text segmentation, brand semantic cleaning, audio and video noise reduction and principal component compression, a content generation network and a communication effect prediction module under brand constraints are established, which is used to realize the integrated calculation of short video generation, brand recognition and market feedback evaluation. The experimental results show that the semantic consistency of the generated content reaches 0.87, the brand information recognition accuracy is 90.8%, the user interaction composite index is increased from 0.71 to 0.81, and the brand communication effect improvement rate is increased from 12.4% to 21.7%. Although the training time was increased from 126 min to 139 min, the inference speed reached 26.4 frame·s⁻¹, which was still feasible for deployment. The results show that AIGC not only improves the production efficiency of brand short video, but also enhances the controllability of brand expression and communication revenue.
