Logo Logo
Hilfe
Hilfe
Switch Language to English

Silva, Samuel; Cunha, Conceição; Teixeira, Antonio; Joseph, Arun und Frahm, Jens (2020): Towards Automatic Determination of Critical Gestures for European Portuguese Sounds. In: Computational Processing of the Portuguese Language, Propor 2020, Bd. 12037: S. 3-12

Volltext auf 'Open Access LMU' nicht verfügbar.

Abstract

Technologies, such as electromagnetic midsagittal articulography (EMA) and real-time magnetic resonance (RT-MRI), can contribute to improve our understanding of the static and dynamic aspects of speech, namely by providing information regarding which articulators are essential (critical) in producing specific sounds and how (gestures). Previous work has successfully demonstrated the possibility to determine critical articulators considering vocal tract data obtained from RT-MRI. However, these works have adopted a conservative approach by considering vocal tract representations analogous to the flash points obtained with EMA data, i.e., landmarks fixed over the articulators, e.g., tongue. To move towards a data-driven method able to determine gestural scores, e.g., driving articulatory speech synthesis, one important step is to move into a representation aligned with Articulatory Phonology and Task Dynamics. This article advances towards this goal by exploring critical articulators determination considering a vocal tract representation aligned with this framework is adopted and presents first results considering 50Hz RTMRI data for two speakers of European Portuguese.

Dokument bearbeiten Dokument bearbeiten