Currently, digital broadcasting tends to personalize the TV watching environment by offering services that can adapt to viewer preferences. In this paper, an intelligent broadcasting system for enhanced personalized-services, based on the semantics of broadcasting contents, is proposed. To implement the system, the MPEG-7 and TV-Anytime Forum (TVAF), as well as agent technology, are employed. For content-level services, real-time content filtering, personalized video skimming, and content-based retrieval using audio characteristic are implemented. To verify the usefulness of the proposed system, we demonstrate it with a test-bed on which content-level personalized services are implemented(1). (c) 2005 Elsevier Ltd. All rights reserved.