In this paper, we have proposed a novel model for wideband speech signal coding, structurally similar to G.711.1 coding system. Unlike G.711.1 coding system, the proposed model has lower complexity that is provided by employing G.711 coders in both branches after applying transformation coding. This way, signal is not being divided into frames, which leads to substantially reduced processing delay. Furthermore, special attention is paid to the determination of optimal values of support range for G.711 coders. System performance is measured by using signal-to-quantization-noise ratio which represents a standard measure of reconstructed signal quality. Experiments are done for male and female speakers, due to difference in speech signal dynamics. The obtained results show that the proposed model ensures experimental gain up to 5 dB compared to G. 711 model, whereas signal processing delay is decreased from 5 ms to 0.25 ms, compared to G. 711.1 standard.

Keywords: G.711 coders, G.711.1 standard, Segmental uniform quantization, Hadamard transform coding