JPS6412658A - Voice detection circuit for dsi equipment - Google Patents

Abstract

PURPOSE:To extract the voice signal without deterioration by detecting a voice while giving variance to a threshold of voice detection of the sender side in response to a voice power at the receiver side so as to avoid mis-detection due to an echo signal generated at the head of speech or the like unable to be erased even by an echo canceller/echo supressor, etc., thereby preventing undesired increase in the line utilization rate. CONSTITUTION:When the calculated power inputted from a 1st voice deciding means 100 is larger than a prescribed value with respect to the calculated power of voice inputted from a 2nd voice deciding means 800 in an adaptive power deciding means 600, the presence of voice is outputted as the result of decision and fed to one input of the selecting means 900. On the other hand, the result of decision of the 1st voice deciding means is fed to the other input of the selection means 900. In a timer 700, a control signal is fed to the selection means 900 by using a voice deciding signal inputted from the 2nd voice deciding means 800 and the input from the adaptive power means 600 is used as the output by the selecting means 900 for a prescribed time. Thus, the signal leaked from the opposite hybrid circuit and unable to be erased even in the echo canceller can be erased. Thus, the undesired increase in the line utilization rate is prevented and the voice signal is extracted without deterioration.