The invention relates to the field of audio signal identification. The invention discloses a method and a system for identifying respiratory sound signals based on visualization, the short-time Fourier transform is used to perform the time-frequency analyze of the cut respiratory sound period signal, and the one-dimensional audio signal is converted into the two-dimensional visual signal. Throughthe image processing and analysis, the data set is formed, and the convolution neural network picture classification is carried out to realize the distinction between normal and three pathological respiratory sounds. Pathological respiratory sound signal murmur is obvious, the murmur formed during exhalation and inhalation has special spectral information. A time-frequency analysis method is used,the short-time Fourier transform is used to perform the time-frequency analyze of the cut respiratory sound period signals, and the one-dimensional audio signal is converted into the two-dimensionalvisual signal. Through the processing and analysis of the images, a data set is formed, and the visual images are classified based on convolution neural network to distinguish normal and three pathological respiratory sounds.