@article{3806,
  author = {Yan Zhengkun, Liu Qian},
  title = {Reinforcement Learning Algorithms to Model English Pronunciation},
  journal = {International Journal of Computational Linguistics Research},
  year = {2023},
  volume = {14},
  number = {3},
  doi = {https://doi.org/10.6025/jcl/2023/14/3/73-81},
  url = {https://www.dline.info/jcl/fulltext/v14n3/jclv14n3_1.pdf},
  abstract = {With the development of artificial intelligence technology, computer-aided pronunciation training systems are gradually receiving attention. However, existing systems have some issues, such as inaccurate judgments and untimely feedback. We propose an improved solution based on deep learning and reinforcement learning to address these issues. This scheme first uses deep learning algorithms to model English pronunciation. Specifically, we used models such as Convolutional Neural Networks (CNN) and Long Short Term Memory Networks (LSTM) to extract and classify features of pronunciation audio. At the same time, we also used reinforcement learning algorithms to automatically adjust training difficulty and feedback methods based on students’ pronunciation to provide more personalized and accurate training effects. Through experimental verification, we found that the improved scheme based on deep learning and reinforcement learning can significantly improve the accuracy and real-time performance of the CAPT system in English pronunciation judgment. This scheme has higher learning efficiency and lower learning costs than traditional pronunciation training systems. In addition, students also showed higher satisfaction and stronger learning motivation towards the program.},
}