METHOD FOR PRODUCTION OF SPEECH REFERENCE TEMPLATES

摘要

Speakers are classified into plural 6 sets of speakers on the basis of the difference in physical features by using speeches which are stable by utterance, the speakers in each classified set are further classified on the basis of the difference in features of speech which occur by utterance, and all of the speech patterns having features of speech which occur by utterance by the speakers belonging to each of the further classified sets are assigned to the same category in the form of speech reference templates belonging to the corresponding category.