Automatic meeting minutes generation using Natural Language processing

In recent years, online meetings have become a significant component of communication in a variety of organizations. To optimize the generation of meeting minutes, we propose a novel approach based on NLP in this work. A model that converts speech to text, summarizes meetings has been proposed in the existing work. The goal of this proposed work is to automate speech recognition, extract key points, summarize the meeting, and extract action items. The approach begins with extracting text from a meeting that is in transcript or audio format. The proposed work utilizes summarization models like BART (Bidirectional Auto-Regressive Transformers), T5 (Text-to-Text Transformer Model), and a Summarization Pipeline. Summarization pipeline model performed well for generating meeting summary. Finally, the proposed work generates concise meeting minutes for the meeting which can be in the form of transcript or Audio. ROUGE Score and Human evaluation will be used to evaluate these models.