The voice quality distinction in Dinka songs

Abstract

The purpose of this work is to study the distinction of the voice quality in Dinka songs, in particular to determine whether and how the phonemic distinction between modal and breathy vowel is conveyed in singing. The methodology adopted consists in the application of two acoustic measurements that have already yielded interesting results on Dinka speech data, namely formant tracking and spectral regression. The outcome of these measurements clearly points towards a neutralisation of the distinction of voice quality in songs. The results of a perception experiment performed on human listeners show more evidence to support the conclusion that no difference in voice quality is conveyed in songs. In the absence of any acoustic cue that could be used by the listeners to detect voice quality, further research is suggested. In particular, the presentation of the same perception experiment to Dinka native speakers could give crucial evidence on the nature of the acoustic cues that may or may not still be present in songs.