7th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2011)

Florence, Italy
August 25-27, 2011

Acoustic Metrics of Vowel Articulation in Parkinson's Disease: Vowel Space Area (VSA) vs. Vowel Articulation Index (VAI)

Shimon Sapir (1), Lorraine O. Ramig (2,3), J. Spielman (2,3), C. Fox (2,3)

(1) Department of Communication Sciences and Disorders, University of Haifa, Haifa, Israel
(2) Department of Speech, Language and Hearing Sciences University of Colorado, Boulder, USA
(3) National Center for Voice and Speech, Denver, CO, USA

Acoustic analysis of speech is a powerful, noninvasive, and cost effective tool to study different aspects of motor speech disorders such as the dysarthria associated with PD. In this presentation we will discuss the rationale for using acoustic analysis, its advantages and disadvantages, and methods to overcome these disadvantages. As an example, we will address the use of vowel space area (VSA) in the study of dysarthric vowel articulation in PD. Although the VSA is theoretically driven, it is highly sensitive to inter-speaker variability, which, statistically speaking, introduces noise. This noise can mask important differences that do exist between speakers with and without PD. Some of this noise can e reduced by logarithmic transformation of the formant frequencies. However, even with this transformation, some statistical noise might be still present. recently sapir and colleagues introduced two acoustic metrics - the vowel articulation index (VAI) and its inverse, the formant centralization ratio (fcr) - that are theoretically driven and empirically tested. These metrics show promise as they effectively reduce inter-speaker variability noise while maintaining high sensitivity to vowel centralization (the latter reflecting abnormally reduced (hypokinetic) articulatory movements in PD). Data will be presented of 38 individuals with parkinson's disease and 14 healthy controls whose speech was effectively differentiated by the VAI, but not the VSA, yet the logarithmically scaled VSA (lnvsa) did significantly differentiate between dysarthric and normal speech, although not as strongly as the VAI.

Index Terms. Parkinson disease, acoustic analysis, speech

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Sapir, Shimon / Ramig, Lorraine O. / Spielman, J. / Fox, C. (2011): "Acoustic metrics of vowel articulation in Parkinson's disease: vowel space area (VSA) vs. vowel articulation index (VAI)", In MAVEBA-2011, 173-175.