Using Audio Description Text for Shot-by-Shot Indexing of Films
James M Turner, Université de Montréal
Suzanne Mathieu, Université de Montréal
The E-inclusion Research Network (<http://e-inclusion.crim.ca/?q=en>) has a goal of “creating powerful audio-video tools… to improve the richness of the multi-media experience for the blind, the deaf, the hard of hearing, and the hard of seeing”. Project 3.1 of the research network involves identification of types of information needed by the visually handicapped to understand moving images. By analysing the audio description provided in a number of films, we identified the types of information described for the visually handicapped, and developed a classification of these types. We analysed the text of the audio description of individual shots, as well as that of user descriptions of the shots. By comparing the two, we can estimate the possibility of automatically deriving indexing to individual shots in a film. Indexing individual shots greatly increases the possibilities for studying films, but it is too expensive to produce such indexing other than automatically. By “recycling” the keywords in the audio description text as indexing terms, access to films at the shot level can be provided.
James M Turner is a professor at the École de bibliothéconomie et des sciences de l’information, Université de Montréal. He holds a PhD in information science from the University of Toronto. His research activities are focused on storage and retrieval of still and moving images, indexing images, metadata for digital images in a networked environment, preservation of digital images, and audio description. He teaches in the areas of multimedia information systems, managing visual and sound information and moving image archives, and preservation of digital information. He is a lifetime member of AMIA and a member of the Section on Audiovisual and Multimedia of IFLA. More information about his professional activities can be found at mapageweb.umontreal.ca/turner.