Content based Multimedia Retrieval using Automatic Speech Recognition

Abstract

Our system deals with retrieval of suitable video based on the user input and voice commands. The microphone takes user voice as input and processes it to convert it into the text. It further checks the repository database of videos to find a video that matches the spoken keyword by the user. A list of keywords is available for the user in order to aid him in searching and querying process. After the relevant video is played as per user’s wish, it is further given an option if the user wants to navigate to particular topic within the video. The system can do the same for the user. Our system basically encompasses and covers application in lecture video domain. The videos in system are based on a wide variety of topics. In general, navigation in videos is too time consuming as it performed by trial and error. However, with the help of this system, searching becomes faster and response time increases. Proper indexed query handling in database makes navigation easier and efficient. The system is now restricted to lecture videos but can be extended to various different domains too in industry. General Terms Fast retrieval, content based retrieval, repository of videos