Design of English teaching speech recognition system based on LSTM network and feature extraction

The development of educational informatization has had a profound impact on college English education. In order to better meet the needs of students, many universities have begun to explore new English education models. This paper designs a speech recognition system for English teaching based on LSTM network and feature extraction. LSTM network is used to process speech signals, extract speech features and then classify them. Feature extraction can transform speech signal into feature vector which can be used by classifier. Using a speech data set containing multiple English pronunciation words, the data set was tested and evaluated. The experimental results show that the speech recognition system based on LSTM network can effectively recognize the differences between different words and can achieve a higher classification accuracy under different noise and interference conditions. The English teaching speech recognition system based on LSTM network and feature extraction has great application prospects. By using the speech recognition system, teachers can evaluate students' English pronunciation level more efficiently and provide personalized teaching programs for students. At the same time, students can also practice English pronunciation independently through the speech recognition system to improve their language expression ability and listening comprehension.