Various multimedia document systems that capture, integrate, and synchronously present media have been implemented in many areas. As data volume grows and the diversity of new medium increases, how to efficiently locate the segments of interest is an important issue. To address this problem, this paper presents a framework that facilitates multiple levels of access to navigated hypermedia documents in three different granularities: HTML slide level, navigation event level, and sentence level. In contrast to traditional approaches, hierarchical access structure of navigated hypermedia documents can be exploited by analyzing temporal meta-information captured in the recording stage or computed in the speech-text alignment process. The proposed framework has been applied in the developed “Web-based Synchronized Multimedia Lecture System” for efficient content access.