The paper introduces an audio rendering system based on the binaural approach, which allows a real-time simulation of spatially distributed sound sources and, in addition to that, near-to-head sources in room-mounted virtual environments. We have been developing a dynamic crosstalk cancellation, allowing the listener to freely move around without wearing any headphones. The paper gives a comprehensive description of the system, concentrating on the dual dynamic crosstalk cancellation and aspects of the integration of a real-time room acoustical simulation. Finally, two applications are described to show the wide applicability of the system.