AI and IoT-based training and evaluation system for college spoken English

Spoken English proficiency is essential for academic and professional communication however, learners often have limited opportunities for interactive practice and consistent feedback. Existing digital spoken English assessment systems lack adaptive evaluation mechanisms and fail to effectively handle diverse acoustic conditions and contextual learning environments, reducing assessment reliability. Therefore, this research aims to develop an Artificial Intelligence (AI) and Internet of Things (IoT)-based training and evaluation system (IoT-TES) for accurate, adaptive, and intelligent spoken English assessment among college learners. Speech data are obtained from a publicly available Kaggle dataset using IoT-enabled microphones, mobile devices, and classroom sensors to capture diverse accents and environmental conditions. Pre-processing includes Wiener filtering for noise reduction, voice activity detection for speech segmentation, and normalization for signal consistency, followed by mel-frequency cepstral coefficients (MFCCs) for feature extraction. The novel contribution of this research is the proposed archimedes optimization algorithm-tuned random forest-extreme gradient boost (AOA-RXGBoost) model, which simultaneously performs optimal feature selection and hyperparameter tuning while integrating random forest for robust prediction and XGBoost for iterative error refinement, thereby improving evaluation accuracy and computational efficiency. The system is implemented using Python for effective model development and validation. Experimental results demonstrate that the proposed model achieves 95.47% accuracy, 95.12% precision, 94.86% recall, and 94.99% F1-score, consistently outperforming existing methods. The developed IoT-TES platform provides real-time feedback, personalized learning support, and adaptive evaluation, offering a scalable, reliable, and intelligent solution for spoken English training across diverse educational environments.