A deep-learning-based speech recognition and real-time feedback system for English pronunciation training in vocational education

In recent years, deep learning (DL) technology has achieved significant results in the field of automatic speech recognition (ASR), with better performance than traditional machine learning (ML) methods. Accurate English pronunciation is the core foundation of oral communication. However, due to the lack of a real English language environment and excessive emphasis on developing reading and writing skills, non-native English learners often have a lower level of oral proficiency. In response to this issue, this article proposes a vocational English oral training system based on DL technology, which can achieve ASR and real-time feedback functions. The system accurately recognizes students' spoken pronunciation through DL models and automatically evaluates the accuracy of pronunciation, providing personalized learning support for students. The results show that the system exhibits high accuracy in identifying English oral pronunciation, and can provide students with immediate and effective feedback to help them correct pronunciation errors in a timely manner and improve their oral expression ability. The application of this system not only makes up for the lack of personalized guidance in traditional teaching, but also provides innovative technical support for vocational English oral teaching.