Crosslingual Audio Information Retrieval Development

Abstract : The Air Force Research Laboratory's Speech and Communication Research, Engineering, Analysis, and Modeling (SCREAM) Laboratory has a commercially-available system to encode, index, archive, and search multimedia events such as news broadcasts. The system is from a company that was formerly called Virage, which is now owned by a company called Autonomy. The Virage system contains a media encoder called a VideoLogger, and it has an audio indexing system from a company called BBN. The BBN audio indexing system gives the SCREAM Laboratory the capability to extract various metadata from audio and/or video content. This report discusses the development of a Virage Media Analysis Plug-ins (MAPs) to allow for translating text generated by the automatic speech recognition system (ASR) as well as a plug-in that allows other ASR or audio processing systems to be integrated with the Virage system.