In this paper, we proposed a new method of signal processing techniques for the application of analyzing four typical lung-sound signals (tracheal breath sounds, vesicular breath sounds, inspiratory and expiratory stridor, and stridor). First, we focus on the estimation method of the sound source number in a lung disease. Second, we calculate the time interval of each sound source signal accurately. Finally, we put the delay time and the number of sound sources as the characteristics, and study the relationships between each pathological lung sound and the characteristics. The innovation of this paper is that utilizing the cepstrum analysis as a new idea to study the pathological diagno-sis of lung diseases.