To provide an inexpensive digital voice processor capable of allowing a digital voice signal inputted to the receiving side to surely coincide with a finally requested emphasis state while surely discriminating whether the format of the inputted digital voice signal is processed by preemphasis processing or not and capable of reducing the cost and the deterioration of characteristics.

The digital voice processor is provided with a preemphasis information discrimination circuit 1 for detecting the preemphasis information of a digital voice signal and outputting an on or off signal as a discrimination signal in accordance with the detected information, a deemphasis filter 2 and an emphasis filter 3, an emphasis on/off setting switch 4 for setting the finally required emphasis ON/OFF state of the digital voice signal and outputting an on or off state signal as an instruction signal in accordance with the set state, and a selector 5 for outputting any one of output signals from filters 2, 3 and the through digital voice signal on the basis of the instruction signal and the discrimination signal.