Skip to main content

Table 3 Recognition accuracy (%) achieved by various approaches for Aurora-2 clean-condition training task averaged across the SNRs between 0 and 20 dB, where AVG (%) and RR (%) are the averaged accuracy rate and the relative error rate reduction over the baseline

From: Enhancing the magnitude spectrum of speech features for robust speech recognition

Method

Set A

Set B

Set C

AVG

RR

MFCC baseline

59.24

56.37

67.53

59.75

–

SSBerouti

69.76

70.47

69.39

69.97

25.40

WFPSNR

71.78

73.66

70.37

72.25

31.05

MMSE log-STSA

72.71

73.58

71.99

72.91

32.71

MSE

77.76

79.89

69.42

76.94

42.72

MVN

73.81

75.02

75.08

74.55

36.77

SSBerouti+MVN

65.71

70.39

66.94

67.83

20.07

WFPSNR+MVN

67.33

70.26

67.35

68.50

21.75

MMSE log-STSA+MVN

73.55

75.67

73.33

74.35

36.28

MSE+MVN

81.85

82.15

76.23

80.85

52.42

HEQ

81.42

83.34

81.51

82.21

55.80

SSBerouti+HEQ

73.04

76.99

73.52

74.71

37.18

WFPSNR+HEQ

74.95

77.30

74.92

75.88

40.08

MMSE log-STSA+HEQ

79.13

80.81

77.99

79.58

49.26

MSE+HEQ

84.19

83.20

78.20

83.80

59.75

MVA

78.15

79.17

79.12

78.75

47.21

SSBerouti+MVA

71.07

75.05

72.05

72.86

32.56

WFPSNR+MVA

69.02

71.70

69.01

70.09

25.69

MMSE log-STSA+MVA

74.39

76.79

74.50

75.37

38.81

MSE+MVA

83.58

85.02

80.58

82.37

56.20