By analyzing the sound field with the impulse responses recorded by closely located four point microphone method, we can get the coordinates, delay times, powers and waveforms information of each reflected sound source. The head-related transfer function (HRTF) of the recorded sound field is then calculated and constructed by the information of reflected sound sources and the general HRTF datasets. Then the HRTF of the recorded sound field can be convolved with any audio file, so that the binaural sound of that recorded field can be reproduced.