The amount of digital material in video lecture archives is growing rapidly, causing the search&retrieval process to be time-consuming and almost impractical. Indeed, after the search, students receive a list of videos and often must use VCR-like functions to find the specific piece of video that covers the searched topic. Therefore, a more efficient method for video retrieval in digital video lecture archives is needed. In this paper, we propose VLB (Video Lecture Browsing), a system designed to facilitate both the retrieval of video lectures within video archives and the finding of the most appropriate segment of a video lecture that covers a searched topic by automatically producing a general picture of the contents of a video lecture. To achieve these goals, the system introduces the idea of timed tag-clouds, which are produced with a combination of aural and visual analysis. Results of a MOS evaluation show that users highly appreciate the timed tag-clouds approach and a comparison study against other popular approaches shows that 93 % of users prefer to use VLB to handle video lectures.