VOICE RECOGNITION DEVICE AND LEARNING METHOD AND LEARNING DEVICE TO BE USED IN THE SAME DEVICE AND RECORDING MEDIUM PROGRAMMED AND RECORDED WITH THE SAME METHOD

摘要

PROBLEM TO BE SOLVED: To realize the enhancing of a voice recognition performance under a noisy environment without limiting the position of a speaker and without increasing the scale of a device and to enable this device to be easily applied even to the conventional voice recognition device. SOLUTION: The feature vector of a voice which is gatherer in a second microphone 101 whose sound receiving sensitivity is secured in a frequency band broader than that of a first microphone 112 is estimated by quantizing a voice input pattern gathered in the first microphone 112 into a voctor while using a first code book 119 and by outputting its index and by selecting the feature vector which corresponds to the index and is stored in a second code book 110 and by connecting them. Moreover, voice parameters stored in a voice dictionary are successively applied as patterns for reference and then both patterns collated in this device.