An Oral English Training System using Manta Ray Foraging Optimization with Xception Model

Speaking English fluency is a requirement for English majors that has to be accomplished in order to gradually complete the course. However, the traditional oral English education strategy is limited in teaching methods and mostly relies on classroom teachers' guidance and observation. Therefore, this research has suggested the Manta Ray Foraging Optimization with Xception model (MRFO-Xception) to improve oral English training system performance using the web application. Typically, this research involves corpus acquisition in order to compile the data required to provide efficient performances. The Hidden Markov Model (HMM) is utilized to preprocess the data from the databases. Then, feature extraction using Mel frequency cepstrum coefficient is utilized to extract the training features based on system characteristics. Hybridization MRFO-Xception is utilized to optimize the features for exact identification with the goal of improving spoken English instruction. The results demonstrated that the suggested MRFO-Xception model, which attained an accuracy of 99.65%, which is better performances than existing models which includes Multifeatured Fusion (METE-MF), Structure of Observed Learning Outcome (SOLO), and BP Deep Neural Network Multimedia (BP-DNNM) approach.