The cultivation of oral expression ability requires Mandarin training, and Mandarin training is an auxiliary tool for students’ oral expression, and the study of Mandarin should be fully utilized to enhance students’ oral expression ability in teaching practice. This paper explores the application of artificial intelligence technology in the teaching of Mandarin oral expression, constructs a Mandarin oral pronunciation assessment corpus based on speech recognition under artificial intelligence technology, and applies the corpus in the actual students’ Mandarin oral expression training. The Mandarin spoken corpus, after the filtering process of the speech signal, is calculated using the MFCC parameter to complete the speech static feature recognition. On this basis, DCT frequency transformation is introduced to describe the dynamic characteristics of speech through differential parameters, and the processed speech signal is output using HMM model. In the analysis of the effect of intelligent spoken Mandarin teaching, the average score interval of the 10 indexes evaluation of the subject class is between [6.53,7], which is much higher than that of the control class. After intelligent teaching, the average scores of answering questions, conversational communication and independent expression of the subject class are 2.978, 3.336 and 2.945 respectively, which are higher than the results of the oral test before teaching, and the students’ oral expression ability is further improved.