Intelligent Detection Method of Spoken English Mispronunciation Based on Machine Learning

In oral English testing, to improve the intelligent detection ability of oral pronunciation errors, an intelligent detection method of oral English pronunciation errors based on machine learning is proposed. According to the specific design principles, a number of parallel processing instructions are established, and then the design and application of the machine learning algorithm are completed according to the recommendation matrix decomposition standard. On this basis, this paper analyzes the acoustic characteristics of oral English mispronunciation, defines four detection indexes: sentence-level TCPP, phoneme-level TCPP, word-level TMPP and phoneme-level TMPP, and completes the adaptive processing of the acoustic model. The example analysis results show that compared with the statistical pattern recognition strategy, the machine learning detection method has less difference between the recognition results and the correct phoneme pronunciation and can more accurately detect the wrong phonemes in the oral English test.