| 摘要 |
PROBLEM TO BE SOLVED: To provide a noise gate for outputting a voice signal in which only stationary noise is removed, without degrading uttered voice of a speaker. SOLUTION: A sound collecting device 1 includes a FFT processing section 11, a noise gate 12 and an IFFT processing section 13. The sound collecting device 1 transforms a collected voice signal NE'T to a frequency spectrum NE'N in the FFT processing section 11. The sound collecting device 1 estimates the noise spectrum N'N of the stationary noise on the basis of the frequency spectrum NE'N of the voice signal in the noise gate 12. When the signal level ratio of the frequency spectrum NE'N and the noise spectrum N'N of the sound voice signal is a threshold or more in the noise gate 12, the sound collecting device 1 outputs the frequency spectrum NE'N as it is, and when it is less than the threshold, it outputs NE'N by reducing a gain. The sound collection device 1 outputs a voice signal CO'T created by inverse transformation of the frequency spectrum CO'N after removing the stationary noise N'N in the IFFT processing section 13. COPYRIGHT: (C)2010,JPO&INPIT |
