UTTERANCE VERIFICATION USING WORD BASED MINIMUM VERIFICATION ERROR TRAINING FOR RECOGNIZING A KEYWORD STRING

摘要

A speech recognition method and apparatus which has a first stage to provide keyword hypotheses and a second stage to provide testing of those hypotheses by utterance verification. The utterance verification used has three separate m odels for each word: one keyword verification model, one misrecognition verification m odel, and one non-keyword verification model. Further, all three are developed independently of the recognizer keyword models. Because of this independence , the three verification models can be iteratively trained using existing speech d ata bases to jointly provide a minimum amount of verification errors.