PURPOSE: To prevent the transmission of an unpleasant voice signal by providing the output of a signal whose voice power is equivalent to zero when a digital voice signal from an A/D converter is less than a prescribed value so as to discriminate a noise level.

CONSTITUTION: A voice is given to a microphone 1 at a transmission station side, and an analog signal is converted into a digital signal by an A/D converter section 2. The voice framed power of the converted voice signal is calculated by a power calculation section 12a in a voice power calculation section 12 by, e.g. selfcorrelation calculation or the like. A noise level comparator section 13 compares the calculated voice frame power with a threshold level prepared in advance and when the power is less than the threshold level, it is discriminated to be a noise level. Furthermore, when the power is equal to or more than the threshold level, the digital voice signal is outputted as it is. On the other hand, a time division multiplex connection timing control section 6 and a control section 7 at a receiver side provide a coded digital voice signal to a digital voice signal processing section 5, where the signal is decoded and the result is given to a D/A converter 4 and the converted analog signal is outputted from a speaker 3.