In a data compression system, a digital signal comprising a series of digital samples and a sampling datum associated with each digital sample is received by a decoder. The sampling datum indicates the sampling interval of the associated digital sample. The decoder includes a microcomputer for storing the digital signal into a memory (M2) and reading each digital sample and the associated sampling datum. The digital sample is divided by the sampling datum to derive a quotient which indicates the slope of the signal to be recovered. The quotient is integrated by an integrator (6b) to provide interpolation between successive sampling points, so that the original signal is approximated by a plurality of line segments.