Aiming at the problem of voiced speech separation in monaural speech separation,  we propose an accurate pitch period estimation method. Firstly, using the shortterm stability of speech and the continuity of the pitch period as clues, we use the cepstrum peak of speech signals to form the pitch spectrum, and the pitch period track is automatically extracted. Then, the spectrum of each harmonic is picked up by using the property that the harmonic frequency is an integer multiple of the fundamental frequency. Finally, the voiced speech is reconstructed by the inverse Fourier transform. Experimental results show that this method can accurately extract the pitch period track and effectively separate voiced signals.