Automatic video programming using LSA and audio transcript

Abstract

In this paper, we present a new approach of automatic video programming (or editing). Our approach uses text processing, statistic analysis and information retrieval techniques. Using Latency Semantic Analysis (LSA) method, audio transcription texts and constraints of the composed video duration, our algorithm selects the video sequences and chains them to create automatically a new video. Our system allows the user to create a video composition by setting its duration.