Communications engineering. An audio compression technique for digital transmission developed around 1970 in the Bell Laboratories. It is defined in the standards ITU-T G.726, G.711 and G.722 and is used for example in DECT.
This method transmits the difference between a predicted value and the signal as a 4- or 8-bit word. The prediction value is calculated from the preceding sampling values. The sampling step width is modified according to the fluctuation of the signal, so that a huge amplitude fluctuation leads to smaller steps and a small fluctuation to wider steps.