The invention features the following: using a calling model to simulate the call interaction between both calling sides; calculating the state of call process at each time point; according to the sound activation detection technology, deciding if said state generates the traffic of network voice call; at the time point of generating the traffic, recording the offset between the time point and the initial time, the message sequence number and the size of the message corresponding to the code to form a traffic sequence; using the interaction between two test ends to determine the start time and the message actually-sending time, and finally generating the network voice call test traffic based on calling model.