Automated Extraction of Meeting Summaries and Action Items Using Whisper and LLMs

Meetings generate large volumes of unstructured conversational data, making manual documentation time-consuming and error-prone. Although platforms such as Google Meet provide automated note-taking and summarization features, these capabilities are typically limited to their own ecosystems and offer restricted customization and export flexibility. This paper presents a platform-independent automated meeting documentation system that converts spoken conversations into structured summaries and actionable insights. The system utilizes Whisper for accurate speech-to-text transcription and Large Language Models (LLMs) deployed locally using Ollama for extracting key discussion points, decisions, and action items. Unlike existing solutions, the proposed system supports multiple input sources including recordings, transcripts, and meetings from platforms such as Microsoft Teams and Zoom.