The use of acoustic parameters for the evaluation of voice therapy of dysphonic patients

This study describes the amount of intra-subject variability to be expected when using an objective, acoustic analysis to evaluate voice. It also considers how such variability affects the assessment of a therapy outcome. In total, 82 patients suffering from chronic dysphonia were asked to produce three sustained vowels /a:/ at a comfortable pitch and loudness level. Three recordings were made over a six-week period. For each sample, acoustic variables characterizing jitter, shimmer and harmonic-to-noise ratio were computed. Statistical analysis was then used to estimate the expected intra-subject variability per acoustic parameter. The variability of perturbation parameters increases with the magnitude of the parameter. Therefore, the inaccuracy was characterized by a relative error (coefficient of variation). For the acoustic parameters, distinctive inaccuracies in series of consecutive sustained vowels were found: the coefficients of variation ranged from 14% to 33%. When an individual therapy is evaluated by acoustic perturbation measures, the ratio of the post-therapy to the pre-therapy value generally had to be below 0.7 to 0.5 to consider the improvement significant.