VOICE RECOGNITION DEVICE, CAR NAVIGATION DEVICE AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To improve a voice recognition rate for a voice of an unspecified speaker.SOLUTION: A voice recognition device includes: a processing section for dividing an input voice in frame units; a processing section for calculating a sound feature quantity of the input voice by frame unit; a processing section which calculates a difference in the sound feature quantity between continuous two frames and outputs the difference as a voice waveform difference; a storage section for storing the difference of the sound feature quantity calculated for a combination of two phonemes including the same phoneme as a voice waveform difference model; a processing section which compares the voice waveform difference calculated from the input voice with the voice waveform difference model, and determines the combination of phonemes having the smallest difference; and a processing section for determining a phoneme expressed by the input voice on the basis of the determined combination of phonemes.