An audio processing method includes: after the terminal accesses the control server, the control server obtains audio capabilities of the terminal through capability negotiation; and the control server selects coded audio data for a terminal and forwards the selected coded audio data to the terminal according to its audio capabilities, without first decoding the selected coded audio data. The audio data does not need to undergo an operation of audio coding and decoding every time when the audio data passes through a control server, and the coding and decoding operations performed by the control server are reduced drastically.