The variable sidetone system provides an enhanced method for inducing the subscriber maintain the amplitude of their voice at an appropriate level by exaggerating the amplitude characteristics of the sidetone returned to the subscriber. Rather than have the sidetone level linked to the amplitude of the subscriber's voice signal in a linear non-adjustable 1:1 relationship (as is done in the prior art), the present variable sidetone system implements an adjustable relationship between the amplitude of the subscriber's voice signal and sidetone levels, thereby allowing the variable sidetone system to provide the necessary acoustic feedback to subscribers to maintain the subscriber's voice signal within the predetermined bounds of the voice encoder.