| 摘要 |
PROBLEM TO BE SOLVED: To easily improve the SN ratio and sound quality in feeling of hearing by converting spectral information of output signals of a spectrum addition means into timedomain signals. SOLUTION: A Fourier transform part 33 processes an output of a voice/ unvoice judgment part 32 with Fourier transform, and analyzing the spectral composition and outputting spectral information; a noise spectrum estimation part 35 estimates noise spectrum included in the output by referring to a spectrum memory 36. An amplitude spectrum subtraction part 34 subtracts the estimated noise spectrum multiplied by a subtraction coefficient from the output of the Fourier transform part 33. When the subtraction coefficient is not suitable, a negative amplitude calculation part 37 makes the amplitude spectrum subtraction part 34 and a subtraction coefficient setting part 38 repeat the processing. A subtraction coefficient setting part 38 sets a subtraction coefficient and outputs to the amplitude spectrum subtraction part 34. And, an inverse- Fourier transformation part 40 which is a time-domain signal transformation means processes an output of a half-wave rectifier 39 with an inverse-Fourier transformation. |
