An algorithm applied to incoming speech before compression and storage so that the amplitude of the stored speech is kept at a target level.