Jonathan Grudin, Anoop Gupta, Liwei He, and Elizabeth Sanocki
As streaming audio-video technology becomes widespread, there is a dramatic increase in the amount of multimedia content available on the net. Users face a new challenge: How to examine large amounts of multimedia content quickly. One technique that can enable quick overview of multimedia is video summaries; that is, a shorter version assembled by picking important segments from the original. We evaluate three techniques for automatic creation of summaries for online audio-video presentations. These techniques exploit information in the audio signal (e.g., pitch and pause information), knowledge of slide transition points in the presentation, and information about access patterns of previous users. We report a user study that compares automatically generated summaries that are 20%-25% the length of full presentations to author generated summaries. Users learn from the computer-generated summaries, although less than from authors’ summaries. They initially find computer-generated summaries less coherent, but quickly grow accustomed to them.