Foreign language speech recognition and correction system aided by machine learning algorithm
Jinmei Zhou, Shuang Su, Rongxing Li · 2024
This paper discusses the application of machine learning algorithm in foreign language speech recognition and correction system, aiming at improving the oral pronunciation quality of foreign language learners. Firstly, the limitations of traditional foreign language teaching mode in pronunciation correction are analyzed, and then the construction scheme of foreign language speech recognition and correction system based on machine learning is put forward. The system constructs an efficient and accurate foreign language speech recognition model through data preprocessing, feature extraction, model training optimization and other steps. In the data preprocessing stage, spectral subtraction is used for noise reduction, and sliding window method is used for framing. In the feature extraction stage, mel-frequency cepstral coefficients (MFCC) is used as the speech feature, and MFCC features are extracted through pre-emphasis, windowing, fast Fourier transform, Mel filtering, logarithmic operation and discrete cosine transform. We chose deep neural network (DNN) as the base model and designed a hybrid model based on long short-term memory network (LSTM) and convolutional neural network (CNN) (LSTM-CNN-DNN model) to improve recognition accuracy. At the same time, L2 regularization term, pre-training, data enhancement and Adam algorithm are introduced to prevent the model from over-fitting and accelerate the training process. In addition, this paper also puts forward foreign language pronunciation correction strategies, including the construction of standard pronunciation database, instant feedback, personalized training plan, interactive exercises, intelligent counseling services and multimodal strategies. The system implementation and test results show that the system can accurately identify and correct pronunciation errors, has good stability, and can meet the needs of a large number of users for speech recognition and correction. This study has successfully constructed an efficient and accurate foreign language speech recognition and correction system, which provides a real-time and personalized pronunciation assistant tool for foreign language learners and is expected to significantly improve their oral English.