ABSTRACT
Meetings, which are crucial platforms for communication and making decisions. The need for efficient summarizing tools has grown as more and more meetings are being videotaped and transcriptions are becoming available. However, summarizing long conversations is difficult, especially when it comes to answering specific questions. To tackle this, we present a multi-domain meeting summary task that is query-based and requires methods for choosing and condensing pertinent meeting segments. We assess multiple strong baselines for summarization, demonstrating the intricacies involved in this undertaking and emphasizing the need for additional investigation. In response, we suggest a chatbot-powered meeting summary system that is trained on the BART model. The system enables users an easy-to-use interface for uploading meeting recordings, interacting with the chatbot, and retrieving important data. It does this by smoothly integrating several technologies, including HTML for interface, JavaScript for dynamic interaction, moviePy for video processing, AssemblyAI audio processing, PyTorch for machine learning, and Fast API for Backend requests. Our research highlights the significance of developing meeting summary methods and provides a workable way to improve efficiency and retention of knowledge in group settings
