A method is provided for synchronizing playback of audio an/or video by a plurality of separate devices in a computer network. Each separate device generates a virtual clock in response to a timing of the audio codec of a received audio stream. Either the virtual clock is generated directly in response to the tick counter of the audio codec, or by a periodic measurement of the timing of the audio codec interpolated by a monotonic clock. A sample rate converter may be used to slightly adjust the frequency of the virtual clock.