Automatic and Data Driven Pitch Contour Manipulation with Functional Data Analysis

Creating stimuli for perceptual experiments in intonation research involves manipulation of pitch contours extracted from spoken utterances. Difficulties arise when changes in the contour shape need to be applied globally and smoothly in the whole pitch curve. Moreover, it is hard to relate a gradual modification in some contour trait to its perceptual counterpart. In this paper we propose a novel approach to stimuli manipulation that is based on an extension of Principal Component Analysis (PCA). Starting from a corpus of pitch curves a parametric description of the principal variation in the curve set is obtained. This allows to locate clusters in this parameter space that are related to linguistic categories. The search for pitch curves with desired perceptual characteristics is carried out by choosing convenient point locations with respect to the relevant clusters. We illustrate this approach in a case study on question/statement opposition in Neapolitan Italian.