| 摘要 |
PROBLEM TO BE SOLVED: To efficiently save power consumption and to prevent an unnatural impression from imposing a remote end talker by providing an output of a voice detection flag based on comparison between a voice detection threshold level and a level of a residual signal so as to provide detection information of a near end voice with a small error. SOLUTION: A voice detection threshold level TH(k) changes with a change in a value resulting from subtracting an echo cancellation amount D(k) from a reception signal level Lr(k). Even when the value resulting from subtracting the echo cancellation amount D(k) from the reception signal level Lr(k) is high, a residual signal level Lu(k) does not exceed a voice detection threshold level TH(k) and a voice detection flag FLG remains zero. That is, near end voice detection is made not from a level Lw(k) of an attenuated transmission signal being an output of an attenuator 10 but from the level Lu(k) of the residual signal being an input to the attenuator 10. Thus, it is avoided that a change in the level Lw(k) of the attenuated transmission signal due to the operation of the attenuator 10 is mis-detected to be a near end voice. |
