摘要 |
PROBLEM TO BE SOLVED: To provide a speech recognition device and method capable of suppressing deterioration of convenience while improving recognition performance. SOLUTION: The speech recognition device 1 includes a microphone 10 for inputting the voice of figures continuously spoken by a user; and a speech recognition unit 20 which recognizes the voice of the continuous figures input through the microphone 10. The speech recognition unit 20 includes a chunk discrimination part 24 which divides the voice of the continuous figures input through the microphone 10 based on meta speech information; a digit number pattern memory 25 which stores the digit number of each divided voice according to the type of speech object; and a recognition unit 27 which recognizes voice by comparing the voice divided by the chunk discrimination part 24 with the digit number of each divided voice stored in the digit number pattern memory 25. COPYRIGHT: (C)2010,JPO&INPIT
|