Acoustic cues for vowel nasalization: A simulation study

Abstract

The manner in which vowel spectra are modified as the degree of the nasal coupling is varied was investigated by means of a simulation of the vocal-tract system. Connecting a side cavity, which represents the nasal sinuses, to the main nasal tract, speech synthesis by simulation could produce naturally sounding nasalized vowels [S. Maeda, “The role of the sinus cavities in the production of nasal vowels,” Proc. IEEE ICASSP82, Paris, 911–914 (1982)]. Such spectral variations were studied in detail to shed some light on acoustic cues for vowel nasalization. The advantage of the simulation approach is that the acoustic characteristics of the nasalized vowels can be calculated for arbitrary coupling magnitude, with a realistic vocal-tract model. The computed vocal-tract transfer functions were then converted to pseudo-auditory excitation patterns. A consistent effect of the coupling, for the 11 French vowels that were investigated, was a flattening of the excitation patterns, extending in the frequency range from 300 to 2500 Hz. Since those vowels with flattened spectra were actually heard as nasalized, it may be suggested that the principal cue for nasalization is spectral flattening in that wide frequency range.