Speech Prosody 2008

Campinas, Brazil
May 6-9, 2008

On the Comparison of Catalan-Spanish Intonation Systems using Statistical Corpus Modeling and Objective Metrics

David Escudero-Mancebo (1), Valentin Cardenoso-Payo (1), Antonio Bonafonte (2)

(1) Departament of Computer Science University of Valladolid, Spain
(2) Department of Signal Theory and Telecommunications, Polythecnic University of Catalonia, Spain

This communication presents an ongoing research on the definition of amethodology to compare the intonation of two different corpora. The two corpora that we compare here, try to be representative of the Spanish and Catalan intonation respectively. As a consequence, the comparison reported here, projects the most relevant differences between the Catalan and Spanish intonation systems. First we model the intonation of the corpora using the MEMOInt methodology and we confront the models obtained by confronting the F0 patterns that share its prosodic function. An objective metric guides the identification of the most distant F0 patterns. The differences can be visualized and listened in perceptual tests. Finally we discuss about the weakness of this ongoing research and about future applications.

Full Paper

Bibliographic reference.  Escudero-Mancebo, David / Cardenoso-Payo, Valentin / Bonafonte, Antonio (2008): "On the comparison of catalan-Spanish intonation systems using statistical corpus modeling and objective metrics", In SP-2008, 563-566.