| 摘要 |
PURPOSE:To recognize a voice with as high precision regardless of the variance of circumstances by determining a reproducing level for feedback of the input voice to a speaker in accordance with the ambient noise level and amplifying the input voice in accordance with the reproducing level. CONSTITUTION:A voice signal is inputted to a voice input part 101, and the ambient noise is inputted to a noise input part 102, and a noise level measuring part 104 receives the input voice from a signal line 112 to measure the noise level and calculates an average noise level for a preliminarily determined time. A control part 105 receives the average noise level from a signal line 113 to determine a reproducing gain. A voice amplifying part 106 receives the reproducing gain from a signal line 114 and amplifies the voice signal received from a signal line 115 with the designated reproducing gain and outputs the result to a voice reproducing part 103 through a signal line 116. Thus, spectrum deformation under the noise is suppressed, and the voice is recognized with a high precision regardless of noise. |
