Decomposition and Recognition of English Speech Features Based on Neutrosophic Set Fuzzy Control and Random Matrix Theory

In order to improve the effect of special decomposition and recognition of English speech, based on the idea of neutrosophic set fuzzy control, this paper uses Bayesian method as the basic algorithm of speech recognition to improve the algorithm in combination with English waveform characteristics. Moreover, this paper uses a semi-supervised learning method to process English speech waveform data, collects relevant data through the English speech input system, and then labels the data and obtains a new English speech data set through training and learning. In addition, this paper uses multiple iterations of labeling to obtain the ideal output data, uses neutrosophic set fuzzy control algorithms and machine learning algorithms to perform English speech feature decomposition and recognition, and uses feature parameter extraction methods to perform signal feature extraction. Finally, this article combines the needs of English speech recognition to build a system model and uses simulation tests to perform performance analysis of English speech feature decomposition and recognition model. The results of the research show that the improved algorithm and system model proposed in this paper have relatively good effects.