Abstract: Speech perception research is one of the main research contents of phonetics. The purpose of this research is to understand the learner's perception of Mandarin vowels and consonants, and to provide a reference for Chinese teaching, broaden the application field of speech learning models and reduce the blindness of teaching. According to the theoretical model of second language acquisition, this paper uses experimental phonetics and statistical methods and designs identification experiments and distinction experiments to study the perception abilities of vowel, consonant, plosive, fricative and affricative for the 20 Uyghur college students who are at the advanced and junior levels of Putonghua respectively. The identification experiment examines the learner's perceived response time and perceived accuracy of vowels and consonants. In the distinction experiment, the spectral distances of vowel pairs, the spectral distances of consonant pairs and their duration differences are calculated to analyze the learner's distinction ability. The experimental results show that the ability perceiving vowels and consonants of the advanced level learners is significantly higher than that of the primary level learners; the learner's response time to the vowel at the back of the pronunciation is fast and the perception accuracy is high, however, the response time to the vowel at the front of the pronunciation is slow and the perception accuracy is low; in addition, the recognition accuracy of the affricative is high, and the lowest is for the fricative; the differences in the spectral distance of vowel pair and in the duration of consonant pair affect the learner's distinction ability, but the spectral distance of consonant pair is not closely related to distinction.