ABSTRACT
Chinese has increasingly become a global language. Compared with the popular international languages such as English, French, and Spanish, traditional classroom teaching methods are often used in the promotion of Chinese. In particular, oral pronunciation requires face-to-face guidance from teachers, which is not conducive to students’ self-study. In response to this problem, this paper proposes to use the most widely used deep convolutional neural network in artificial intelligence technology to judge the standard of the spoken pronunciation of non-Chinese students of native language, and according to the “Mandarin Test Grade Standard” jointly issued by the National Language and Writing Committee and other departments The Chinese phonetic standards in, automatically grade the correctness of students’ spoken pronunciation. The activation function of the deep convolutional neural network(DCNN) uses the ReLU function, the loss function uses the cross-entropy cost function, and the training algorithm for the DCNN is also given here. Practical experiments on 100 speech fragments sent by 10 students show that the proposed DCNN is feasible and effective.
