Application of Artificial Intelligence and Speech Recognition Technology in English Listening and Speaking Ability Assessment System

This paper, with an innovative pronunciation correction system for spoken English, grounded in speech recognition technology, has been conceived and designed. The core of this investigative work revolves around two principal domains: speech feature extraction and the development of correction methodologies. The study endeavors to delve into the nuances of English speech recognition, with a particular focus on the application of the Hidden Markov Model (HMM). Central to the system's functionality is the accurate identification and rectification of pronunciation errors. A speech recognition framework based on Sphinx-4 is proposed. At the same time, a computer-aided scoring system for oral English test is designed and implemented. It can realize the recognition of incorrect sounds through the error operation of the result of the tag. Voice control is performed sequentially according to the feedback path. The "advanced control group" was used for speech correction. Experiments show that the algorithm can recognize and correct speech effectively. This method can suppress the noise very well. It has a good recognition ability for various emotions, and the recognition accuracy of English accent reaches 90%. The experimental results show that the designed controller has a high response time.