Table 1 Comparison of proposed scheme with existing work under clean condition

Method of [39]

78.07%

88.19%

96.60%

84.35%

97.64%

SEC-MLP using paralinguistic feature

Without data augmentation

62.95%

39.72%

79.26%

57.36%

65.18%

With data augmentation

83.97%

69.07%

91.28%

79.52%

85.75%

SEC-SVM using paralinguistic feature

83.74%

90.71%

96.13%

99.03%

96.11%

SEC-MLP using spectral feature

Without data augmentation

77.12%

57.57%

88.90%

75.75%

77.67%

With data augmentation

89.39%

78.38%

94.50%

86.84%

90.42%

SEC-SVM using spectral feature

86.73%

95.42%

99.41%

98.43%

98.26%