Differential companding PCM by using optimum differential bit
Hiroki Uchiyama, Wasaku Yamada, M. Suzuki · Electronics and Communications in Japan (Part III Fundamental Electronic Science) · 1989
Abstract Two speech coding methods combining dif ferential PCM (DPCM) with near instantaneous companding (NIC) are proposed in this paper. In general, if NIC is applied simply to DPCM, then the waveform cannot be reconstructed in the decoding process because of the bit removal during compression. to solve this problem, a method has been proposed which is based on the accumulation of the removed bits, but the companding ratio in this case is not sufficient for low bit rates. In Method I proposed in this paper the compressed differential data obtained by NIC of the differential data are modified to minimize the error between the original data and the data decoded sequentially, sample by sample, for the given number of compressed bits. Method II is similar to Method I, but an optimum scale factor is chosen from three adjacent values. Thus, the error power is minimized over the block. Segmental SNR is used as a quality measure, and the results of the computer simulation show that higher speech quality is obtained than in the case of first‐order predictive ADPCM (ADPCMf). The hardware realization of the proposed methods is extremely simple because both coding and decoding require merely additions and subtractions. This article presents the principle of the methods and the results of the simulation.