The paper deals with the selection of the test signal for the voice quality measurement (VQM) in telecommunication networks. It lays the fundamentals to use the optimized multi-sine as test signal in place of both the natural and the artificial voices, suggested by the ITU-T recommendations. The effectiveness of the optimized multi-sine to the VQM is investigated by using the four algorithms: (i) perceptual speech quality measurement, (ii) perceptual speech quality measurement plus, (iii) perceptual evaluation of speech quality, and (iv) perceptual analysis measurement system. Tests are performed in order to compare the mean output index of the four algorithms feed by the optimized multi-sine, the natural and the artificial voice signals, respectively. In order to take into account the effects of the actual telecommunication networks, different characteriztic noise is used to corrupt both the multi-sine and the voices: (i) modulated noise reference unit, (ii) impulsive noise, and (iii) additive white Gaussian noise. Further tests are performed by simulating both the transmission and the reception processing of the GSM standard. The tests show the effectiveness of the technique based on the optimized multi-sine to the VQM with respect to the traditional one.