Multimodal Saliency and Fusion for Movie Summarization Based on Aural, Visual, and Textual Attention | Synapse