Sixth International Conference on Spoken Language Processing
In this paper we discuss the differences of the utterance verification/rejection between speaker-dependent and speaker-independent speech recognition. A general guideline will be given to handle the verification task for the two type speech recognitions. We also discuss some specific methods have been commonly deployed in various speech recognition systems. In this study we describe two example approaches, an online score approach for speaker-dependent and a filler-cohort modeling approach for speaker-independent speech recognition. From our experiments we conclude that there is no common approach of confidence measure estimation for both SD and SI speech recognition.
Bibliographic reference. Zhang, Yaxin (2000): "Utterance verification/rejection for speaker-dependent and speaker-independent speech recognition", In ICSLP-2000, vol.4, 218-221.