Ourmuse: Plot-Specific AI Music Generation for Video Advertising

The integration of music and video is a pivotal aspect of creating impactful advertisements. This study explores the application of Artificial Intelligence (AI) in generating music tailored for video advertisements through a system we developed, named Ourmuse. By leveraging advanced Generative AI models, Ourmuse aims to enhance the synchronization between music and video content. The methodology involves extracting and analyzing still images and texture context from the video. A critical aspect of our approach is the plot-based training of the AI model by identifying transition points between video scenes, where the video is segmented into distinct parts rather than being treated as a single unit. This segmentation ensures that Ourmuse produces music that is coherent and contextually appropriate for each segment, enhancing the overall impact and cohesion of the final advertisement. By focusing on multi-modal inputs and plot-specific training, this research aims to develop a robust AI model capable of revolutionizing the music composition process for video advertisements.