Voice-Activated Meeting Assistant for Real-Time Agenda Tracking and Summarization

Meeting management plays a vital role in organizational productivity, yet challenges such as agenda drift, time wastage, and inefficient note-taking often hinder effective discussions. Traditional meeting assistants primarily perform transcription but lack contextual understanding, making them incapable of managing dynamic meeting flows. To address these issues, this paper introduces VAMAR (Voice-based Adaptive Meeting Assistant for Real-time), a context-aware system designed to ensure real-time adaptive agenda tracking and on-the-fly summarization. VAMAR integrates advanced speech recognition with a transformer-based intent detection model to interpret conversational context, detect topic shifts, and adapt agendas accordingly. The Adaptive Agenda Model (AAM) dynamically restructures discussion points by merging or dividing topics, while the Participant Engagement Module (PEM) minimizes agenda deviation through subtle, non-intrusive voice prompts. Simultaneously, the Real-Time Summarization Engine (RTSE) generates concise, context-sensitive summaries and action points accessible during the meeting. Evaluation results demonstrate over 90% accuracy in agenda tracking and high user satisfaction due to improved focus and reduced interruptions. By combining real-time understanding, adaptability, and automation, VAMAR represents a next-generation intelligent meeting management framework that enhances collaboration efficiency and decision-making quality.

Ҳали таржима қилинмаган