Capturing important information from an audio conversation

Abstract

A conversation between people can involve exchange of important information. Often in an ongoing conversation, someone has to manually keep on taking notes of important points. This is a tedious process and can involve errors. In today's era of technology, there is a need for an automated process for extracting important information from a conversation, with minimal manual efforts. In this paper, we propose methods of transcribing a conversation between two people. We will consider two major modes of conversation-when the conversation is happening over a phone call, and when the conversation is happening in person. In addition to transcribing a conversation, this paper will also suggest ways to extract important parts of a conversation. We will extract important information from a conversation, using three different approaches-noun phrase extraction, named entity extraction and open information extraction method. Since current mobile operating systems provide limited support for transcribing a phone call, we will suggest ways of transcribing a call, and extracting important information from it. © 2017 IEEE.