Minutes of Meeting Generation for Online Meetings Using NLP & ML Techniques

Within the realm of remote work, meetings serve as unique spaces where individuals summon to exchange concepts and reach consequential conclusions. Currently, they are written down by hand, and while it can be challenging to keep up, occasionally they overlook crucial information. Therefore, automating the process of creating meeting minutes with machine learning and natural language processing techniques has emerged as a promising way to do rid of the tedious human processes involved in recording and summarizing meeting discussions. This study presents a novel method for automating the creation of meeting minutes through the use of machine learning (ML), natural language processing (NLP) algorithms, and several summarization models, including Google's PEGASUS, BART, LED, Meeting Summary, and bartconversation summary. In order to verify the model's functionality, audio clips from different websites are gathered, and the results are computed and compared using the ROUGE-1, ROUGE-2, ROUGE-P, ROUGE-R, and ROUGE-F scores. With the aid of the Whisper AI model and pyAnnote, the suggested solution makes extensive use of natural language processing (NLP) to evaluate meeting transcripts. It is specifically made to create transcripts for online meetings, starting with the addition of time stamps, speaker identification, and the assignment of each sentence to the appropriate speaker. After that, the text is divided up into distinct sentences for processing. This technique accurately identifies and summarizes important themes, crucial choices, action items, and other significant details discussed throughout the discussion.