Also known as voice activity detection (VAD). A voice compression mechanism based on the fact that there are predictable periods of inactivity in normal human speech. During those pauses and other periods of silence, the pulse code modulation (PCM) signals are halted through a technique known as silence suppression. On the receiving end of the transmission, a matching algorithm reconstitutes the PCM signals, reinserting the periods of silence. As DSI works on the basis of statistical probabilities, it is employed effectively only when there are a significant number of voice conversations supported. For example, DSI applied to 96 channels of PCM-encoded voice yields compression of 2:1, which reduces the PCM bandwidth requirement of 64 kbps to only 32 kbps, on average. DSI builds on the earlier analog technique of time-assignment speech interpolation (TASI). See also bandwidth, compression, encode, PCM, and TASI.