In this paper we present preliminary results of a novel unsupervised approach for high- precision detection and correction of errors in the output of automatic speech recognition sys- tems. We model the likely contexts of all words in an ASR system vocabulary by performing a lexical co-occurrence analysis using a large corpus of output from the speech system. We then identify regions in the data that contain likely contexts for a given query word. Finally, we detect words or sequences of words in the contextual regions that are unlikely to appear in the context and that are phonetically similar to the query word. Initial experiments indicate that this technique can produce high-precision targeted detection and correction of misrecog- nized query words.

Click Here to View the Paper In XML Format