PURPOSE: To efficiently attain a voice response by automatically judging whether the media of the other party side terminal equipment is a telephone set, facsimile, or MODEM by a corresponding responding part, canceling a circulating echo in a hybrid circuit, and exactly receiving a signal from the other party terminal equipment.

CONSTITUTION: Whether the media of the other party side terminal equipment is telephone sets 1-1-1-k, facsimiles FAX 2-1-2-m, or MODEMS 3-1-3-n is automatically judged by a voice responding part 7, FAX responding part 8, and MODEM responding part 9 in a voice responding device. At the same time, when a push button input or a voice input signal is not inputted, a circulating signal estimating circuit 52 of a network control part 5 prepares an approximate echo (a) approximating to the circulating echo in the hybrid circuit 51 of the transmission voice signal from the transmission signal from the, responding part 7, and controls the approximating echo (a) according to an erased residual echo (b) as that the cancelled residual of the circulating echo can be the minimum.