| 摘要 |
PURPOSE:To realize a high recognition rate with regard to the speaker adaptive voice recognizing device for recognizing a voice of a new speaker. CONSTITUTION:The device is provided with plural templates 20 for managing the corresponding relation of the voice feature quantity and the voice contents, a converting part 21 for converting the voice feature quantity managed by the templates 20 in accordance with a parameter, a leaning part 23 for leaning the parameter by which the voice feature quantity of the templates 20 converted by the converting part 21 comes to coincide with the voice feature quantity of a learning voice input, and setting it to the converting part 21, and a selecting part 25 for selecting the template 20 for showing that which is near the voice feature quantity of a selection voice input by comparing the voice feature quantity of the selection voice input and the voice feature quantity converted by the converting part 21. This device is constituted so that the voice contents of a new speaker voice are recognized by converting the voice feature quantity of the template 20 selected by the selecting part 25 and generating a speaker template 26 for a new speaker. |
