Pre-processing of fundamental frequency contours of speech for automatic parameter extraction

While it is quite straightforward to obtain a natural F/sub 0/ contour from a set of parameters of a model for F/sub 0/ contour generation, the inverse problem cannot be solved analytically. This paper presents a method for pre-processing a measured F/sub 0/ contour to obtain its approximation consisting of third-order polynomial segments that are continuous and differentiable everywhere. It is shown that the proposed method allows one to obtain first-order approximations to the parameters of accent commands for about 90% of all the accent commands.