Objective voice analysis in dysphonic patients: New data including nonlinear measurements

Objectives/Hypothesis: This report describes a comparative study of objective voice evaluation using a multiparametric protocol including aerodynamic parameters and linear and nonlinear acoustic parameters recorded on an EVA (R) workstation and perceptual voice analysis by a jury. Study Design: A total of 449 samples were retrospectively selected including 391 patients with pathological voices (308 women and 141 men) and 58 controls with normal voices (38 women and 20 men). A prospective complementary study concerning 43 female patients and 3 controls is presented. Methods: Objective measures included fundamental frequency (Fo), intensity, jitter, signal-to-noise ratio (SNR), Lyapunov coefficient (Lya), oral airflow (OAF), estimated subglottic pressure (ESGP), maximum phonatory time (MPT) and vocal range. A jury of 4 experienced listeners was instructed to classify voice samples (continuous speech) according to the G (overall dysphonia) component of the GRBAS score on a Visual Analogue Scale (VAS) secondarily transformed in a scale ranging from 0 for normal to 3 for severe dysphonia. Results: It was shown that a nonlinear combination of only 7 parameters in women ( vocal range, Lya, ESGP, MPT, OAF, SNR, and Fo) and 6 parameters in men (vocal range, Lya, OAF, ESGP, Fo, SNR) allowed classification of 81% voice samples in the same grade as the jury in women and 84% in men. In the prospective study, 80.5% were correctly classified with the same set of objective measurements. Discussion: The relative importance of the different objective parameters in this type of discriminant analysis is dealt with. Special emphasis is placed on Lya. Copyright (c) 2007 S. Karger AG, Basel.