In our recent work on speech analysis and resynthesis, we have been using predictor (LPC) derived area functions to describe the spectrum of the acoustic signal. It has been found that small changes in the value of these parameters do not affect the speech quality. If the acoustic signal is indeed overspecified by these parameters, then bit reduction is achieved by specifying the data less often, at fixed or varied intervals, or with less accuracy; however, we have chosen a scheme where less frequent sampling of the parameters is determined by the properties of the speech mechanism. We have found that if the boundaries of the steady‐state portion of the phonemes are found, the steady‐state portions, as well as the transitions between the phonemes, can be represented by straight lines. This method allows for the description of the acoustic signal with two sets of points per phoneme. Numerous sentences have been encoded by this method, and the resulting sentences do not sound differently from the sentences from which the data were derived. Such a scheme could be used for a rule‐synthesis scheme, as well as for segmentation of speech in speech recognition schemes.Subject Classification: [43]70.50, [43]70.40.