The invention provides an audio synthesis processing method and system. The method comprises the steps of: collecting voice data of a plurality of participants of current frames; calculating the difference between the voice data of the current frame of each participant and the voice data of a previous frame, and generating the voice difference data of the current frame of each participant; according to the number of the participants, setting an attenuation level and a sign bit for the voice difference data of the current frame of each participant; and accumulating the voice difference data of the current frames of the participants after the attenuation levels and sign bits are set, then adding the accumulated value to mixing data of previous frames, and generating mixing data of the current frames. According to the invention, the background noise can be effectively eliminated, and the direct-current bias in the voice data is also effectively eliminated; in addition, the algorithm related to the invention is simple, the control is simple, fewer resources are occupied, found mixing of as many as 64 participants can be simply realized, and the algorithm can be conveniently realized in low cost sound mixing designs.