Detecting Pronunciation Errors in Spoken English Tests Based on Multifeature Fusion Algorithm

In this study, multidimensional feature extraction is performed on the U-language recordings of the test takers, and these features are evaluated separately, with five categories of features: pronunciation, fluency, vocabulary, grammar, and semantics. A deep neural network model is constructed to mo...