Abstract

Developers of high-quality text-to-speech systems are increasingly concerned with voice source control rules. In order to derive such rules from natural speech, the parameters of a source model must be derived from the acoustic signal. This is done by parameterizing the results of glottal inverse filtering. In this paper we compare a number of different inverse filtering procedures and the ease with which their results can be parameterized. It is shown that Closed Glottis Interval Covariance LPC is equally powerful as more sophisticated techniques, because it is the only known method that can strictly be limited to the closed glottis interval.