Generating Relevant Question Answer from Video

Online video lectures are currently a widely used form of instruction as online education has grown significantly. The process of creating question-and-answer pairs from these videos could be laborious, time-consuming, and exhausting. In this study, we offer a cutting-edge methodology for creating questions with meaningful answers, which would simplify the laborious process of question development. To extract text from videos, we employ speech recognition software. By sending a synchronous request to Google Speech Recognition, the video is converted to text. Two times, first for extractive summarization and then to perfect it for question creation, we employed the T5_base transformer. The condensed text is utilised to produce keywords, which are then supplied to the T5_base transformer with SQuAD dataset tuning to produce questions, along with context. In order to examine the calibre of the queries, we also compare our model to a few previously developed models using the same input. We discover that when compared to other models, our model produces superior questions and responses.