Interactive Video Virtual Assistant Framework with Retrieval Augmented Generation for E-Learning

In the era of a new digital world, there has been a decline in social interactions among humans in a physical setting. These concerns have made educational institutions focus on holistic development, which is recognized to be crucial for the overall growth and performance of a student. Nevertheless, concerns like inadequate student teacher and peer interactions, in an online setting can have adverse effects in the long run. This paper presents an analysis and the benefits of utilizing an interactive virtual assistant framework as a means to abridge the growing disconnect of modern technologies from the traditional classroom setting, closing in on the gap brought about by virtual learning medium. This research study examines the online and hybrid learning approach, highlighting its drawbacks in comparison to conventional teaching methodologies while trying to improve the current situation. The virtual assistant framework presented in this study makes use of cutting-edge technologies in the field of NLP dialogue systems and video generation using Wav2Lip, (a lip-sync model) and GFPGAN model for face-specific enhancement. The Virtual assistant is developed and demonstrated as a learning aid in both hybrid and physical classrooms. Further applications and performance improvements of the framework are discussed.