TRANSCRIPTION AND TEXT SUMMARIZATION USING DEEP LEARNING TECHNIQUES

Large amount of data are generated every day in the Internet. There is a need for more efficient techniques for effectively obtaining crucial information from the large data generated. One method for locating the most important and useful information in a document is text summarization. It condenses it into a shorter form while preserving the primary idea. Voice-powered transcription reduces the time by taking care of the preliminary transcribing in real-time. The efficiency is improved as the generated summary has proper context and the most important information. Voice recognition techniques help in generating the document quickly. The system is designed to use audio inputs from the user to convert it into textual format. This text along with a compression ratio is provided as an input to a model that summarizes the text using a combination of the algorithms. Sentences are graded based on the weights. The most important lines that maintain the critical information are generated as the text summary. The model used for summarization is made up of multiple sub-modules that assign the scores for all the sentences in an exclusive manner. These scores are later aggregated to find the most important lines which are then displayed to the user.