Variable bit rate bandwidth compression transmission through a network

Abstract

Digital samples of an analog speech/music signal are converted to compressed digital samples from which first and second bit assignment signals are derived to respectively represent quantization levels corresponding to high and low transmission rates. The compressed digital samples are quantized according to the first bit assignment signal and split into higher significant bits corresponding to the second bit assignment signal and lower significant bits corresponding to the difference between the two bit assignment signals. Supplementary information is combined with the higher significant bits and assembled into a higher-priority cell for transmission to an ATM network and the corresponding lower significant bits are assembled into a lower-priority cell which may be discarded by the network. At the destination, the supplementary information is used to recover the first and second bit assignment signals. Using the recovered second bit assignment signal and the difference between the two bit assignment signals, the higher significant bits are combined with the corresponding lower significant bits when the lower-priority cell is present and combined with synthesized lower significant bits when the lower-priority cell is absent. The combined bits are dequantized according to the recovered first bit assignment signal and decompressed to recover the original digital samples.