Perception of vowel sequence with varying speaker size

Abstract

Speech sounds convey information about the size of the speaker. Several studies have demonstrated that human vowel recognition is possible even for an unnatural size range, and have revealed that size factor normalization can be achieved automatically in the auditory system. In this study, we further investigated the characteristics of the size normalization process, using vowel sequences with temporal changes in the speaker size. In the current experiments, listeners were presented with six-vowel sequences in which the vocal-tract length was alternated vowel by vowel. The experimental results for the identification of the vowel sequence showed that it was increasingly difficult for listeners to identify vowels in the correct order as size alternation was applied with a higher speed and to a larger degree. However, they showed the high performance of vowel recognition when serial order judgment between vowels was not required, and in this case the performance deterioration caused by size alternation became small. The observed deterioration of sequence identification is likely to have been caused not by a failure in size normalization in the auditory system but because of a difficulty in judging the serial order between vowels in the sequence with rapid size changes. The results suggest that the auditory system has a fast process for normalizing speaker-size information and that it operates appropriately even when a sequence contains the temporal alternation of vocal-tract length.