Table 2.

UAR(%) Performance comparison of the proposed method with the state-of-the-art methods on the URTIC dataset

Model	UAR(%)
ComParE functionals + SVM (Schuller et al.) [33]	64.00
ComParE BoAW + SVM (Schuller et al.) [33]	64.20
VOI + SVM (Huckvale and Beke) [21]	66.34
VOW + SVM (Huckvale and Beke) [21]	66.47
MOD + SVM (Huckvale and Beke) [21]	67.95
GPPS + SVM (Huckvale and Beke) [21]	66.07
MFCC + GMM (Cai et al.) [7]	64.80
CQCC + GMM (Cai et al.) [7]	65.40
PSP + SVM (Suresh et al.) [35]	64.00
VMD + SVM (Deb et al.) [15]	66.84
MFCC + Autoencoder (Kao et al.) [25]	65.81
eGeMAPS + NN (Teixeira et al.) [37]	66.90
MFCC + FV + PCA + SVM (Vicente et al.) [17]	64.92
Vowel-like regions MFCC + DNN [43]	61.93
Proposed (MFCC + LPC + SMOTE–Tomek links + DNN)	67.71