Key points are not available for this paper at this time.
Abstract Delta modulation with syllabic companding is one of the voice digitalization systems for which experimental studies have been conducted to evaluate speech quality. First, appropriate circuit constants are provided by use of the objective evaluation measure S/N O for the experiments and then studies are made on the features of the quantization noise in the system. Second, the cepstrum distance measure obtained from linear prediction analysis is introduced as an objective evaluation measure and its relationship to subjective evaluation is also studied. The following conclusions are made. (1) Lowering the sampling frequency to 32 kHz has little effect on articulation and can provide fair or better than fair quality if appropriate signal input levels and received speech volumes are maintained. (2) The effect of circuit noise at about 2.0 mV at the encoder input on voice articulation is deemed to be equivalent to having the same circuit noise at the decoder output. (3) The detectability threshold of transmission errors is about 2.4 × 10 −4 , which proves that the system is less prone to code errors than the log‐PCM system. In addition, the coincidence of the cepstrum distance measure with the subjective evaluation value is better than that of the segmental SNR.
Itoh et al. (Mon,) studied this question.