Comparative Analysis of Models to Improve Technical Speaking through AI

Mazhar, Ahmed Zeeshan (2025) Comparative Analysis of Models to Improve Technical Speaking through AI. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Technical speaking is an essential skill in academic and professional domains, requiring clarity, precision, and structured delivery. Traditional training approaches rely on subjective evaluation and delayed feedback, limiting real time improvement and failing to assess verbal and non-verbal aspects simultaneously. This research proposes an AI based multimodal system that integrates speech and facial emotion analysis to enhance technical speaking performance through real time feedback. The system uses the CMU MOSEI dataset for speech processing and the Roboflow emotion dataset for facial expression detection. Speech analysis employs machine learning models, including Extra Trees and Random Forest Regressors, to predict key linguistic indicators such as Words Per Minute, filler word count, complexity score, and bias detection. For visual analysis, YOLOv8 and YOLOv11 models detect six emotion classes with high precision and real time inference capability. Combined models demonstrate robust performance, accurately predicting speech features and tracking emotional states with confidence scores above 0.80 for dominant expressions such as happy and surprise. Evaluation metrics, including MSE, MAE, R² score for audio and mAP for video, confirm the system’s effectiveness in delivering comprehensive feedback. Findings indicate that the integration of audio and visual cues significantly improves accuracy compared to unimodal systems. This study contributes to both research and practice by introducing an adaptive and efficient solution for real time communication assessment, supporting personalized learning and professional development.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Vamadevan, Arundev UNSPECIFIED
Uncontrolled Keywords:	Technical Speaking; Artificial Intelligence; Natural Language Processing (NLP); Speech Analysis; Facial Expression Recognition
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	20 Aug 2026 09:47
Last Modified:	20 Aug 2026 09:47
URI:	https://norma.ncirl.ie/id/eprint/9562

Actions (login required)

View Item