Lecture Stream: An Intelligent Lecture Summarization System
摘要
This paper presents a solution to address challenges in traditional academic lecture settings, where notes-taking is a time-taking work. Students often face challenges in finding key points due to lots of information and limited accessibility to the lectures for later review. This solution provides students with an invaluable resource, especially those who want to balance their academic and extracurricular activities, by giving accessible summaries of the missed lectures. The lecture summarization system in this paper includes technologies like wireless microphone capture, cloud storage, and natural language processing (NLP) to improve academic learning experience. By seamless recording of the lecture, storing it securely, and accurately transcribing the content, our system can help students with easy and efficient access to summarized information. It provides a user-friendly interface and enhances learning and engagement for students and also sets a new methodology for lecture management and content summarization in educational institutions. We have emphasized student and teacher authentication, access control on the basis of roles, and following data privacy regulations. Leveraging BART model, our system can generate efficient summaries of different lecture lengths, processing audio in chunks. Through rigorous testing, our system achieved transcription accuracy over 95%, with a minimal latency of 100 ms, transcribing a 20-min audio file in 1 min. These outcomes our proposed system demonstrate the system’s robustness, scalability, and efficiency, setting a benchmark in lecture management and content summarization within educational settings.