Fifth ISCA ITRW on Speech Synthesis

June 14-16, 2004
Pittsburgh, PA, USA

Estimating Phrase Curves in the General Superpositional Intonation Model

Jan P. H. van Santen, Taniya Mishra, Esther Klabbers

Center for Spoken Language Understanding, OGI School of Science & Engineering, Oregon Health & Science University, Beaverton, OR, USA

Superpositional intonation models posit that the pitch contour, F0, can be quasi-additively decomposed into component curves such as phrase curves, accent curves, and segmental perturbation curves. Currently, these component curves can only be estimated if one assumes a specific superpositional model, such as the Fujisaki model. A method is proposed for estimating phrase curves that is model-independent, and thus can be used to explore the validity of the general concept of superposition, which component curves are needed, and what the properties are of these curves. Results are presented that show that accurate estimates of phase curves can be obtained for pitch curves generated by the Fujisaki model and by a variant of the Bell Labsí Linear Alignment model, using the same method and parameter settings.

Full Paper

Bibliographic reference.  Santen, Jan P. H. van / Mishra, Taniya / Klabbers, Esther (2004): "Estimating phrase curves in the general superpositional intonation model", In SSW5-2004, 61-66.