The Synchronized Multimedia Integration Language (SMIL) allows Web designers to design complicated and vivid multimedia presentations in a declarative manner. In this paper, an abstract semantic model called Multimedia Augmented Transition Network (MATN) is used to model the conceptual structure of SMIL, as well as the temporal relations and synchronization control. The advantages of using the MATN model are its simplicity and ease of modification (scalability). Users can easily generate their favorite multimedia presentations using MATNs and render them using the SMIL players. Since the SMIL specification is quite new to the Internet society, the functionality of the players is limited, i.e., they often acquire proprietary platforms or cannot deal with MPEG or RTP media objects. Another contribution of this paper is to propose a novel architecture based on Java JMF technology for tackling with such constraints.