In order to evaluate the quality of college English teachers’ oral classroom teaching, this article proposes a method of using artificial intelligence technology to evaluate the quality of oral English classroom teaching, establishes a teaching quality evaluation model based on SOFM neural network, and proposes two types based on deep neural networks. A more noise-robust scoring algorithm for acoustic models. "Listening" is an important way for learners to acquire information, and the process also includes refining and reprocessing the information when acquiring information. "Speaking" is a direct way of information feedback, and the process includes the organization and reconstruction of information. (Goodness of Pronunciation), and the performance is improved by 7.6%.