Disclosed are techniques for transmitting bundles of silence indicator (SID) frames during a voice call among a plurality of access terminals. In an aspect, a source access terminal detects a transition to a silence state, generates, in response to detection of the transition, at least a first bundle of SID frames, wherein each SID frame of the at least the first bundle of SID frames includes data representing comfort noise to be played at one or more target access terminals of the plurality of access terminals during the silence state, and transmits the at least the first bundle of SID frames to a base station serving the source access terminal. In an aspect, the base station receives the at least the first bundle of SID frames, and periodically forwards SID frames of the at least the first bundle of SID frames to the one or more target access terminals.