AUTOMATIC EXTRACTION OF MOD FUNDAMENTAL FREQUENCY CONTOUR

The generation process model of the fundamental frequency contours (F0 contours) of speech is known to be capable of generating F0 contours quite close to observed ones. The extraction of model parameters from an observed contour, however, requires an iterative process starting from a set of initial parameter values. In order to guarantee a rapid con-vergence to an optimum solution, the values should be ap-propriate ones. We already have developed a method of au-tomatically extracting these from given F0 contours, and ap-plied it to Japanese sentences with good results. The method is based on approximating an observed contour by a con-tinuous curve differentiable everywhere. In the present pa-per, it was applied to English utterances. Experiments were conducted for 4 native speakers ’ utterances with 14.5 % and 17.5 % of average miss and false alarm rates for the accent commands, and 35.7 % and 15.5 % for the phrase commands. 1