摘要 |
<P>PROBLEM TO BE SOLVED: To distinguish between a noise component and a voice component in an input signal. <P>SOLUTION: A voice processing device 110 includes: a framing unit 120 that cuts out an input signal by a predetermined frame unit and generates a framing input signal; a spectrum generating unit 122 that converts the framing input signal from a time domain to a frequency domain and generates a spectrum pattern collecting a spectrum for each frequency; a peak detection unit 124 that detects a peak spectrum having a peak from the spectrum pattern; and a harmonic sound determining unit 126 that determines a harmonic sound spectrum being a harmonic structure having a relation of a fundamental sound and a harmonic sound in the peak spectrum. <P>COPYRIGHT: (C)2012,JPO&INPIT |