K. Kim and H. Kim (Korea)
Utterance Verification, Confidence Measure, OOVRejection
Many utterance verification (UV) algorithms have been studied to reject out of vocabulary (OOV) in speech recognition systems. Most of conventional confidence measures for UV algorithms are mainly based on hypothesis or log likelihood ratio test. But these measures take much time to evaluate the alternative hypothesis or anti-model likelihood. The proposed confidence measure that makes use of a momentary best Viterbi score is more time and memory efficient than conventional algorithms because it does not need to model anti-model and to calculate the alternative hypothesis. Using the proposed confidence measure, we can achieve significant performance improvement as well as the reduction of computation time and memory requirement.
Important Links:
Go Back
