The short time feature Mel frequency cepstral coefficients (MFCC) were extracted (for the process, see Fig. 2) from 117 sound samples of Chinese leaf warbler (Phylloscopus yunnanensis), which were collected from 7 regions in Beijing, Gansu, Sichuan and Shaanxi (Table 1, Fig. 1). Then we constructed the cluster tree (Table 2, Fig. 3) of the sound characteristics of different geographic populations by using pattern recognition method. Finally, we discussed the mechanism of geographical variations in songs of Chinese leaf warbler. The results showed that there were no significant differences between MFCC feature of the songs and their distances or habitats (Pearson, r =﹣0.036, P = 0.762, n = 117). However, the sound of Chinese leaf warbler and altitude showed significant differences (Pearson, r =﹣0.836, P < 0.001, n = 117). This is the first attempt for geographic population relationship based on short time feature of sound of birds.