Relationship of intelligibility and subjective quality of digitally coded speech

Abstract

An experiment has been performed to investigate perceptual effects of digital encoding of speech. The aim was to gain information about the influence on intelligibility and subjective quality of three distortions that arise in analog‐to‐digital conversion: bandwidth reduction (B), peak clipping (C), and amplitude quantization (Q). Allowing two levels of each distortion, we produced eight encoders encompassing all possible combinations of B, C, and Q. By means of a consonant recognition rhyme test and a five‐point opinion scale, 12 subjects provided intelligility and subjective quality data about each encoder. The results show that it is not always possible to predict the effects of multiple distortions from information about the effects of individual distortions and that the distortions influence intelligibility and subjective quality differently. For example, with respect to consonant recognition, Q and C reinforce one another; occurring together they cause more recognition errors than the sum of the errors caused by the two distortions occurring individually. On the other hand, the effects of Q and C on subjective quality are essentially additive. Of the three distortions, B has the most serious effect on intelligibility while C has the strongest influence on listener opinions.