摘要 |
PROBLEM TO BE SOLVED: To provide a speech recognizer that is improved in speech recognition accuracy by detecting an utterance segment with high accuracy and segmenting speech data. SOLUTION: An utterance segment detection processing unit 25 includes a speech segment detection part 25a for detecting the utterance segment from the speech data and an image segment detection part 25b for detecting the utterance segment from the image data. The utterance segment is detected by switching and controlling switches SW1, SW2, SW3 and by selectively using either one of the speech data or the image data. A detection processing switch control part 26 controls the switches SW1, SW2, SW3 depending on conditions, and determines which data to be used depending on the conditions. COPYRIGHT: (C)2008,JPO&INPIT
|