Article révisé par les pairs
| Résumé : | We propose a new glottis signal model. Most glottal pulse shape models have been obtained so far by concatenating a small number of curved segments. We propose here an alternative approach which consists of expanding the signal into a combination of a finite set of basic time functions. These functions are chosen taking into account the point-like and non-linear character of the acoustic voice source. We establish a mathematical relationship between the weights of the individual time functions and the Fourier coefficients of the signal being modelled. The control parameters are the frequency and amplitude of a cosinusoidal driving function. The spectral envelope of the output signal (i.e. the glottal pulses) changes with the fundamental frequency, and its spectral content evolves with the amplitude of the driving function. © 1990. |
