Automation of Minutes of Meeting

Project Title

Automation of Minutes of Meeting

Project Area of Specialization Computer ScienceProject Summary

Meetings are essential for sharing information, understanding different perspectives of problems and their solutions, collaborations, assigning tasks, and taking feedback and explanations. One of the challenging aspects of all sessions is to create minutes of the meeting. The advent and advanced voice recording technology and voice to text transcription technology enable us to think of an automatic tool that generates the minutes of the meeting automatically by recording the session. This final year project is a product that automatically makes minutes of the meeting with annotated text from the discussion of the persons involved in the meetings. Minutes of the Meeting (MOM) is an AI-powered voice note application for automated note-taking and summary preparation for in-person meetings. MOM is a voice and document collaboration app that delivers Enterprise Meeting Assistant - EMA MOM’s EMA will join your session and record the meeting, observe the speaker, and extract essential pointers from the attached documents. MOM will require some facts regarding Meeting such as Meeting Chairperson, other attendees, the agenda of meeting to automate the meeting’s summarization. MOM will use these references to generate the Minutes of Meeting which one can share with the concerned persons and add in the record. Apart from summarizing the meeting, MOM clumps the meeting points according to the agenda of the meeting, and one can easily recall the update on a specific program. Also, it will assign tasks to attendees, which will help the manager to take followup. MOM uses speech to text for producing the meeting transcript. National Language Processing and heuristics Techniques for highlighting the critical point from meeting transcript and attached documents, clustering the talk according to agenda and assign tasks to attendees. MOM also guarantees data protection, which is recorded during a meeting.

Project Objectives

• Transcribe audio to text
• Clustering according to agendas
• Classification of tasks
• Extracting important features from the text
• Minutes of the meeting generation

Project Implementation Method

Initially, we will pre-process the recorded audio data for noise reduction, which may hinder the further process. Then, by analyzing the audio data, we will separate the dialogues of a different person and will annotate the audio data. These annotated audios will further be converted into textual data using Speech to text mechanism. The produced separate texts will then be amalgamated sequence wise with assigning randomly generated tags to participants. We will examine the provided metadata to improve the transcript generated. The annotated transcript will be classified into agenda discussion, the suggestion given and task assigned. Later the agenda discussion will be gathered agenda wise using a particular clustering algorithm. We will use more contextual text processing techniques to generate points. In conclusion, a comprehensive Minutes of the Meeting will be the output of this whole pipeline.

Benefits of the Project

Every enterprise, either small or large, organize meetings daily. For instance, in a software house following the agile sort of software management methodology, daily scrum meetings are held to discuss the progress and further plan of action. The minutes of the meeting method is the way to note down the key insights of the meeting, including the previous development and proposals for an agenda. The primary purpose of this project is to not only automate this essential process of enterprise meetings but to present it in an improved manner as compared to human power. This application will tend to provide these minutes of the meeting concerning the terminologies the enterprise is built. The main focus of this application is to minimize the exploitation of human energy in doing an excessive amount of work daily for creating minutes of the meeting.

Technical Details of Final Deliverable

MOM will provide a complete automated solution for the professional world. As a system, audio and video recordings will be given as an input with some metadata supporting the meeting points. For instance, the meeting documents. MOM will also require the agenda of the meeting and its participants. In this manner, the MOM can target generating the minutes concerning the terminologies of the meeting in a contextual way. Furthermore, the instruments to capture audio and video data are to be preinstalled in the meeting room. Whenever an individual starts to speak the system detects it through the gestures and starts to record. This system will be an amalgamation of relevant Computer Vision, Machine Learning, and Natural Language Processing techniques. For instance, to detect who is speaking an efficient Computer Vision Algorithm is required. Similarly, Machine Learning and Natural Language processing techniques will provide near-optimal contextual information. The three mediums of data, namely audio, video, and meeting documents, will validate the results. This data will minimize the inaccuracy of the meeting’s minutes generation process. Moreover, MOM will provide several additional facilities that include speaker wise contribution, highlighting the key terminologies, and sharing the minutes with colleagues in one click. Moreover, the archiving of all minutes from the past, searching on repositories are some of the added features of Minutes of Meeting.

Final Deliverable of the Project Software SystemCore Industry ITOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources