Abstract: A system and method for automatically controlling signal gain and minimizing distortion due to widely varying signal levels are disclosed. The method comprises the steps of calculating an initial gain factor from a source signal and gradually varying the initial gain factor to within a preselected band of allowable values when the source signal is above a preselected minimum level and the initial gain factor is above the preselected band. The circuit comprises means for calculating an initial gain factor from a source signal and means for gradually varying the initial gain factor to within a preselected band of allowable values when the source signal is above a preselected minimum level and the initial gain factor is above the preselected band. The method and circuit are particularly adept at processing audio signals carrying spoken words, the signals characterized by periods of silence between the words.