By differentiating the input audio signal and detecting a point at which the differential value satisfies a predetermined condition as the sample points, the data generated to obtain a discrete amplitude data and timing data indicating a time interval between each sample point in each of the detected sampling points part 3, provided with a correlation computing section (4) for calculating the correlation data with the amplitude data and timing data, for any input speech correlation data with a number of sound generated by the correlation calculation unit (4) with respect to the Inquiring the correlation data stored in the data placed by the memory (6) created in advance as to how and to recognize the input speech.
展开▼
