Lecture videos are becoming ubiquitous medium for e-learning process. E-lecturing has evolved more competent popular lectures. The extent of lecture video data on the World Wide Web is increasing fastly. Therefore, a most appropriate method for retrieving video within huge lecture video library is required. These videos consist of textual information on slides as well as in presenter’s speech. This paper estimates the virtual utility of mechanically recovered text from both of these sources for lecture video retrieval. This approach gives content based video searching method for getting most relevant results. To implement this system, firstly we have to separate out contents on presentation slides and speaker’s speech. For mining textual information written on slides we apply optical character recognition algorithm and to translate speaker’s speech into text we will apply automatic speech recognition algorithm. Finally, we will store extracted textual results into database against particular timestamp and unique id by performing automatic video indexing. When user will put a search query, then results will be displayed according to video contents. This technique will be beneficial for the user to search a suitable video within a short period of time.