Abstract :
Acoustic transmission-line vocal tract models are used to study formant frequencies, bandwidths and amplitudes of hyperbaric heliox speech versus those of speech in air at 1 ATA. The models account for energy losses due to glottal impedance, lip/nostrils radiation, wall vibration, viscous friction and thermal conduction. New wall impedance values are presented, matching measurements of the closed tract resonance. On basis of, a uniform tube model, an extended version of the classic Fant-Lindquist formula [1] describing formant frequency shift is developed, and formulas for bandwidth and amplitude shifts are given. A multitube vocal tract model is applied for analysis of the effects of nonuniform vocal tract cross-sectional area on the formant shift.
Conference_Titel :
Acoustics, Speech, and Signal Processing, IEEE International Conference on ICASSP '85.