| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a voice recognition device which determines fitness of a probability distribution of speech emergence for a topic. <P>SOLUTION: If a distance between the probability distribution of speech emergence at an initial state which is set when starting voice recognition processing and the probability distribution of speech emergence after the initial state is longer than a predetermined distance (S512:Yes), or a topic determination flag which indicates at least once the probability distribution of speech emergence formed status that fits the topic is ON (S514:Yes), the voice recognition device calculates a bias of the probability distribution of speech emergence in use (S522). If the bias of the probability distribution of speech emergence is equal to or larger than a predetermined value (S524:Yes), or the bias of the probability distribution of speech emergence is smaller than the predetermined value (S524:No), and when the probability distribution of speech emergence in use is switched to the other appropriate probability distribution of speech emergence (S528), the voice recognition device informs a navigation side that a voice recognition result is adoptable as a command. <P>COPYRIGHT: (C)2012,JPO&INPIT |
