| 摘要 |
PROBLEM TO BE SOLVED: To provide a speech recognition device capable of extracting a part where a speech recognition error likely occurs, when a speaker moves. SOLUTION: The speech recognition device includes: a microphone 1 for inputting speech uttered by the speaker; a speech recognition section 2 for extracting a letter code string corresponding to utterance from speech input by the microphone 1; a speaker motion information acquiring section 4 for detecting speaker's motion; and a speech recognition correction score calculation section 5 and a correction requirement determination section 6, in which it is determined that, when the speaker's motion detected by the speaker motion information acquiring section 4 is larger than prescribed motion, the letter code string extracted by the speech recognition section 2 from the utterance when the speaker's motion starts, does not meet a content uttered by the speaker. COPYRIGHT: (C)2009,JPO&INPIT
|
