Lightweight class multi-side conference sound mixing method and device

Provided is a lightweight class multi-side conference sound mixing method and device. The method comprises the steps that (1) after a client side uses an AMR encoder for encoding voice, voice PCM data and data length are obtained, the encoded PCM data are subjected to framing processing, each frame voice energy value is computed, the fact that a frame is a voice frame or a non-voice frame is determined according to the frame voice energy value and the data length, and accordingly the probability values of the voice frames in the voice PCM data are obtained in a statistics mode; and (2) a server side selects current voice streams of two speakers with the highest voice probability values according to the received voice probability values, whether the superposition principle is used for carrying out sound mixing on the at most two selected voice streams is determined according to the two voice probability values, and finally a voice packet obtained after sound mixing is transferred. According to the method, the shortcoming that portable equipment such as a mobile phone is weak in computing capacity is ingeniously overcome, meanwhile, the computing amount of a server for sound mixing operation is greatly lowered, and the lightweight class multi-side conference sound mixing method and device can be widely used in a multimedia multi-side conference system.