| 摘要 |
PROBLEM TO BE SOLVED: To improve speech quality in a video conference system or the like. SOLUTION: This processor is provided with a power calculation circuit 30, a background noise calculation circuit 31, a voice detection circuit 32, a delay circuit 33, variable attenuation circuits 34 and 35, an attenuation amount control circuit 36 and an adder 37. The sound or silence of speech voice signals is detected by monitoring the relative level of voice signals for which the power of input voice signals is calculated and background noise. Also, the input voice signals are delayed and the two of the delayed input voice signal and the non- delayed input voice signal are variably controlled based on the detection of the sound or the silence. Further, output voice signals for which the two variably attenuated input voice signals are added are sent out. Thus, even if the background noise is large and the utterance level of a speaker is low, the incorrect detection of the speaker is reduced. Also, speech head omission at a communication initial part at the addition changeover in voice signal detection is prevented and the delay of the voice signals is minimized. |
