Abstract: From the current situation, it can be seen that there are certain deficiencies in the current models of spoken English analysis. In order to improve the English spoken analysis effect, this study builds an English spoken analysis model based on transfer learning and analyzes the performance of spoken English recognition. In order to make full use of the characteristics of speech feature modes to compensate for the shortcomings of single mode in speech recognition, this paper proposes a multimodal shared speech feature learning method, that is, multimodal shared speech feature learning method based on locality, sparsity, and identifiable typical correlation analysis. The method introduces locality, sparsity and discriminability, and the method effectively improves the English spoken recognition effect to a certain extent. In addition, this paper designs a controlled experiment to analyze the performance of the system model. The research results show that the algorithm has certain effects and can be applied to practice.
Keywords: Transfer learning, spoken English, speech recognition, system model, feature analysis
