SPEECH RECOGNIZING APPARATUS AND OPERATING METHOD THEREOF

摘要

PURPOSE: To make it possible to input information used by a user's business application by comparing a constituted partial speech model with speech notations recorded in a vocabulary data base in response to the reception of speaking data. CONSTITUTION: After spoken alphanumeric data is received through a telephone line 31, an automatic speech recognition(ASR) algorithm program 1021 accesses a partial word model data base 1032 and constitutes a model which consists of a series of phonemes and represents the sound of the received spoken words. Then the ASR algorithm program 1021 accesses the vocabulary data base 1031 and compares the constituted series phoneme model with voice representation series recorded in the vocabulary data base 1031. When the constituted series phoneme model matches a voice representation series recorded in the vocabulary data base 103, it is recognized as a recorded word corresponding to the matching voice representation series after the vocalization.