Silence detection for multimedia communication systems

Silence detection and removal is an essential building block of any multime dia video conferencing system. It reduces the bandwidth requirements of the underlying network transport service and helps to maintain an acceptable e nd-to-end delay for audio. We analyze the requirements for a silence detect ion algorithm hosted on a multimedia communication system, and propose a no vel low-complexity algorithm operating in the non-linear mu-law domain. Aft er discussing the constraints which are imposed by the architecture of the system hardware (computer, packet-based network), we show that several rece ntly proposed silence detection algorithms fail to meet all of these constr aints. A new approach is then introduced, based on the small- and large-sig nal behavior of the speech waveform in the mu-law domain. The new algorithm is compared with a recent design that meets several of our requirements; e xperimental results indicate that it performs significantly better in the p articular environment at hand.