PURPOSE: To improve the recognition accuracy by modulating the outputs of a pulse sound source and a noise sound source with the featured frequency of a standard pattern, selecting one of the two sound sources according to data representing the current frequency distribution, and outputting the modulated wave as an audible wave.

CONSTITUTION: The pulse sound source 12 and noise sound source 13 are provided and their outputs are modulated with the featured frequency of the standard pattern. Further, one of the two sound sources 12 and 13 is selected by switching a switch S16 according to the data representing the current frequency distribution and the modulated wave is outputted as the audible wave. Consequently, the standard pattern in a dictionary becomes audible and the voice at the time of registration can be remembered; if the standard pattern is registered together with an unnecessary sound, it can be heard and the recognition accuracy of the device is improved.