AbstractAs a means of information compression in speech communication, the pause‐in‐speech deletion technique has been applied to digital speech interpolation, etc. This paper examines a pause‐in‐speech technique, which, after detection of a voice segment, goes backward some tens of milliseconds before the voice segment and can turn a voice switch on. Used in a delay‐allowable communication system, this permits efficient compression of the pause in speech while maintaining minimum quality degradation and can be a cost‐effective economization method. the possibility of a backward hangover process with pause deletion and rewriting of voice segments in pause segments in the buffering process using addressing control is discussed. Experimental results reveal that the backward hangover process (hangover time of 32 ∼ 64 ms) permits less voice deletion (especially at word fronts, which are essential to quality reproduction) than the ordinary hangover process. the total cost effectiveness is discussed with respect to a voice storage system and it is shown that the cost of the system can be reduced without speech quality degradation by applying the backward hangover process and addressing control at buffering.