Speech is typically modelled using time-domain or frequency-domain simulations of the acoustic field in the vocal tract. Using a biorthogonal modal decomposition, it is shown that time-domain finite-difference simulations can be transformed algebraically into equivalent formant synthesizers, the parameters of which vary in time and are calculated directly from the laws of physics. Examining the structure of the equivalent formant synthesizer, it is observed that formant excitation is largely due to internal modal coupling effects, induced by rapid perturbation of the acoustic eigenmodes caused by vibration of the glottis, and does not rely precisely on external sources provided by boundary conditions. This leads to a novel interpretation and justification of traditional models of the glottal source.
Cite as: Ramsay, G. (1998) An analysis of modal coupling effects during the glottal cycle: formant synthesizers from time-domain finite-difference simulations. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 0670, doi: 10.21437/ICSLP.1998-362
@inproceedings{ramsay98b_icslp,
author={Gordon Ramsay},
title={{An analysis of modal coupling effects during the glottal cycle: formant synthesizers from time-domain finite-difference simulations}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 0670},
doi={10.21437/ICSLP.1998-362}
}