Synchronizing audio and video streams in packet-based networks requires synchronization of packet timestamps. The present invention provides such synchronization without resort to a network time standard. In one embodiment of the present invention, pairs of timestamp synchronized signals, such as audio and video signals, not having a common timestamp clock are mixed. One of the signals, for example, the audio signals, is mixed first while preserving the original audio timestamps. The preserved timestamp information is then used to synchronize the timestamps of the unmixed signals, in this example the video signals, to provide synchronization of all signals. In another embodiment, the present invention uses packets containing calibration of timestamps to reduce jitter. The present invention also includes specifications for a packet for transmitting timestamp information.