| 摘要 |
PROBLEM TO BE SOLVED: To improve convenience by allowing call registration based on smooth speech recognition.SOLUTION: A speech recognition device includes speech processing means 6 and speech recognition means 14. The speech processing means 6 preliminarily registers sound data of each type of operation sounds of an elevator with a recorded sound database in association with operation information indicative of a type of operation of the elevator and, when input sound 18 includes two pieces of waveform data due to simultaneous input of user's speech 16 and operation sound 17 of the elevator to a microphone in an elevator car, acquires elevator operation sound data 19 corresponding to a current operation of the elevator from among sound data of operation sounds of the elevator recorded in the recorded sound database and performs sound processing 20 of subtracting waveform data of the elevator operation sound data 19 from the waveform data of the input sound 18 and outputs processed data 21. The speech recognition means 14 uses the processed data 21 to perform speech recognition. Elevator call registration is performed on the basis of the speech recognition result. |
