Skip to main content
. 2020 Jul 13;29(3):1596–1607. doi: 10.1044/2020_AJSLP-20-00001

Table 4.

Threshold values and performance measures for the Analysis of Dysphonia in Speech and Voice (ADSV)–based cepstral peak prominence (CPP) and Praat-based smoothed CPP (CPPS) classifiers.

Variable ADSV CPP
Praat CPPS
Sustained vowels Rainbow Passage Sustained vowels Rainbow Passage
Threshold 11.46 dB 6.11 dB 14.45 dB 9.33 dB
ROC AUC .91 .95 .93 .98
Accuracy 79.4% 87.7% 77.4% 94.5%
TPR 0.77 0.87 0.74 0.95
FPR 0.08 0.08 0.02 0.10
TNR 0.92 0.92 0.98 0.90
PPV 0.98 0.98 0.99 0.98
Youden's J 0.69 0.79 0.72 0.85

Note. ROC = receiver operating characteristic; AUC = under the ROC curve; TPR = true positive rate; FPR = false positive rate; TNR = true negative rate; PPV = positive predictive value.