Abstract

Voice-Activated Question Answering (VAQA) systems represent the next generation capability for universal access by integrating state-of-theart in question answering Q&A and automatic speech recognition (ASR) in such a way that the performance of the combined system is better than the individual components. This paper presents an implemented VAQA system and describes the techniques that enable the terative refinement of both Q&A and ASR. The results of our experiments show that spoken questions can be processed with surprising accuracy when using our VAQA implementation.

Citations