Stereophonic sound systems – such as two channel stereo or surround – became extremely common. Yet, we don’t always have the chance to listen the music content on such systems, in these cases using a headphones may serve as a good alternative. However, we often ignore the fact, that playing even two channel stereo content without modification on headphones is incorrect method, let alone multi-channel contents. Furthermore, opposed to normal listening circumstances using headphones always causing the sound space moving along with the listener’s head. In the first part of my thesis I summarize the factors and cues which allow us in determining the direction of sound sources based only on hearing. Then, I present the evolution of stereophonic sound systems from historical point of view, then I describe the most remarkable systems in detail too. In the second part of the thesis I present a system that I made to be able to down-mix literally all kinds of stereophonic sound contents to headphones using real-time convolution, with mathematical and physiological models taken in account. The filters are chosen from a database according to the position of the listener’s head and the position of the virtual sound sources. Also there are interpolation and range extrapolation applied on the filters.