English-Speaking Teaching System with Natural Language Processing Technology Based on Artificial Intelligence Algorithms

English speaking is a very important skill of English ability for students. At the same time, English speaking learning needs a demanding environment. Because English teachers have heavy teaching tasks and limited time. This research developed an English-speaking teaching system with natural language processing technology based on artificial intelligence algorithms, which can solve this problem very well. The system engages the most advanced natural language processing technology based on artificial intelligence algorithms to correct and feedback students' oral problems in a timely, fast, and targeted manner, which is beneficial to improve students' speaking learning efficiency and interest. This system proposes an improved Bidirectional Encoder Representations from Transformers model, which is simply called BERT and based on Transformer's bidirectional encoding representation. The accuracy of this algorithm for spoken English recognition is significantly higher than other classical algorithms. This system has been tested and evaluated in some universities, and the learning efficiency and satisfaction of students is significantly higher than other oral English learning systems.