DEVICE FOR SUPPORTING SPEECH HANDICAPPED PERSON'S CONVERSATION, METHOD FOR PREPARING CONVERSION RULE, METHOD FOR SUPPORTING SPEECH HANDICAPPED PERSON'S CONVERSATION, CONVERSION RULE PREPARING PROGRAM AND SPEECH HANDICAPPED PERSON'S CONVERSATION SUPPORTING PROGRAM

摘要

PROBLEM TO BE SOLVED: To utter contents uttered by a speech handicapped person as an ordinary person's voice by automatically synthesizing utterance from other registered utterance models even when the contents are not included in registered utterance models. SOLUTION: The speech handicapped person's conversation supporting device is constituted of a speech feature extraction part for extracting features from an inputted speech signal and outputting an input speech feature, a feature similarity processing part for finding out the similarity of the input speech feature with the speech features of respective original utterance models registered in an original utterance model feature registering part, an utterance model specification part for determining an original utterance model having a speech feature of high similarity with the input speech feature which is found by the feature similarity processing part, a target utterance model determination part for extracting a target utterance model corresponding to the original utterance model from a plurality of target utterance model groups registered in a target utterance model data registering part, and a speech synthesis part for synthesizing the speech of the target utterance model extracted by the target utterance model determination part and outputting the synthesized speech. COPYRIGHT: (C)2005,JPO&NCIPI