Video indexing using speech recognition techniques in audio channel -preliminary system design

Multimedia document archiving is an important and interesting issue with the current growth of multimedia documents. Among the different types of archiving, audio/video indexing plays a crucial role. For audio indexing techniques, most current audio indexing systems use a combination of speech recognition and information retrieval. A large vocabulary continuous speech recognition (LVCSR) system is used to produce time aligned transcripts of the given collection of speech. Information retrieval techniques are then employed on these recognized transcripts to identify locations in the text that are relevant to the search request. In this paper, we introduce the technology details of an audio indexing system for a specific application of the National Library of Medicine.