The ongoing TIWO project is investigating the synthesis of language technologies, like information extraction and corpus-based text analysis, video data modeling and knowledge representation. The aim is to develop a computational account of how video and text can be integrated by representations of narrative in multimedia systems. The multimedia domain is that of film and audio description – an emerging text type that is produced specifically to be informative about the events and objects depicted in film. We suggest that narrative is an important concept for intelligent multimedia knowledge management. We then give an overview of audio description for film and discuss the integration of video and text data in this context.