Analysing Oral Presentation Quality: A Multimodal Approach Incorporating Presentation Delivery and Slide Quality

Abstract

"This research addresses the challenge of enhancing presentation skills through the development of a mobile-responsive web application called 'Presently'. The app aims to help speakers improve pronunciation, vocabulary, emotional expression, and slide quality. It provides users with personalized feedback to enhance their presentation delivery. The methodology employed involves the use of automatic speech recognition (ASR) for pronunciation and vocabulary analysis, audio and video analysis for emotion and tone detection, and computer vision for slide quality assessment. The app utilizes machine learning algorithms to provide users with tailored suggestions for improvement based on their performance. Initial testing of the 'Presently' prototype shows promising results. For face emotion detection component achieved an accuracy rate of 74.2%. Emotion and tone detection in audio analysis reached an accuracy rate of 95.6%. Slide quality assessment using computer vision techniques showed global recognized standards in identifying content accuracy and aesthetics."