FRAMED DIGITAL VOICE SUMMING FOR TELECONFERENCING

In a teleconferencing system activity levels of digital voice signals are compared over activity selection times and the signals are multiplied, over frame delays, by predetermined constants dependent on the relative levels of activity. The most active signal is multiplied by a constant of at least about 0.8, so that signal dominates through a frame delay, but the frame delays are sufficiently short that multiple speakers can be understood from the summed signal. No speaker maintains dominance for a sufficiently long time to substantially suppress speech of others.