Bruce R. Geratt
Marc S. Dauer
Gerald S. Berke
Div. of Head and Neck Surgery, UCLA School of Medicine, CHS 62-132, Los Angeles, CA 90024
VA Med. Ctr., West Los Angeles, CA 90073
Dysphonic voices are commonly analyzed using automated digital voice analysis software. However, reliability and validity of acoustic measures obtained from these programs has been questioned, especially in the assessment of dysphonic voices. This study evaluated the agreement among various commercially available speech analysis programs and a handmarked voice analysis system. Sustained vowels of 25 male and 25 female speakers with mild to severe dysphonia were digitized and jitter, shimmer, and harmonics-to-noise ratio were analyzed using CSpeech, Computerized Speech Laboratory (CSL), and SoundScope. The effect of dysphonic severity on the measures provided by these four programs was also assessed. The clinical utility and theoretical limitations of automated, perturbation-based measures will be discussed, especially with regard to the indeterminant nature of F0 in some types of dyhsphonia.