| 摘要 |
PROBLEM TO BE SOLVED: To improve recognition ratio, to reduce the amount of calculation and to shorten the time reguired for recognition by narrowing down the standard pattern of a monosyllable to be compared by a simple operation. SOLUTION: A sound signal inputted from an input part 2 is analyzed into the time change of a frequency spectrum by an analytical part 4 for every prescribed period and outputted as a feature parameter. A selection part 10 selects either one of the standard pattern of a 1st monosyllable group and that of a 2nd monosyllable group previously stored in 1st and 2nd standard pattern memories 12, 14. A comparing/recognizing part 6 compares the feature parameter of the speech signal with the standard pattern of the selected monosyllable group and outputs the recognized result of a monosyllable corresponding to the nearest standard pattern. |
