Skip to main content

Table 1 Comparison of average PESQ, STOI, and SDR in different languages

From: Supervised Attention Multi-Scale Temporal Convolutional Network for monaural speech enhancement

 

Training dataset

English

Mix

English\(\varvec{\rightarrow }\)Mix

Test dataset

Metrics

PESQ

STOI (%)

SDR

PESQ

STOI (%)

SDR

\(\varvec{\Delta }\)PESQ

\(\varvec{\Delta }\)STOI (%)

\(\varvec{\Delta }\)SDR

English

Unprocessed

1.96

92.6

13.63

1.96

92.6

13.63

-

-

-

CRN

2.41

94.8

18.49

2.45

95.0

18.56

0.04

0.2

0.07

MSTCN

2.68

94.8

16.40

2.66

95.1

16.32

− 0.02

0.3

− 0.08

LSTM-IRM

2.80

95.9

18.84

2.80

96.0

18.92

0.00

0.1

0.08

GCRN

2.76

95.5

19.98

2.76

95.5

19.98

0.00

0.0

0.00

GaGNet

2.89

96.1

20.43

2.92

96.1

20.35

0.03

0.0

− 0.08

Conv-TasNet

3.07

96.6

21.67

2.93

96.2

20.94

− 0.14

− 0.4

− 0.73

DCCRN

3.17

96.6

21.20

3.14

96.5

20.57

− 0.03

− 0.1

− 0.63

DPCRN

3.16

96.6

20.75

3.11

96.4

20.59

− 0.05

− 0.2

− 0.16

SA-MSTCN\(^{1}\)

3.26

96.6

20.11

3.30

96.8

20.57

0.04

0.2

0.46

SA-MSTCN\(^{2}\)

3.29

96.7

20.46

3.35

97.0

21.13

0.06

0.3

0.67

Chinese

Unprocessed

2.31

88.2

16.81

2.31

88.2

16.81

-

-

-

CRN

2.44

86.8

16.77

2.76

89.4

20.30

0.32

2.6

3.53

MSTCN

2.26

86.9

14.94

2.99

91.5

17.45

0.73

4.6

2.51

LSTM-IRM

2.80

90.4

18.37

3.15

92.6

20.97

0.35

2.2

2.60

GCRN

3.09

90.2

21.46

3.09

90.2

21.45

0.00

0.0

− 0.01

GaGNet

2.77

89.1

18.02

3.16

91.2

21.38

0.41

2.1

3.36

Conv-TasNet

2.95

90.5

20.12

3.16

91.1

21.39

0.21

0.6

1.27

DCCRN

2.66

89.4

17.98

3.43

92.7

21.70

0.77

3.3

3.72

DPCRN

3.01

90.7

18.99

3.40

92.5

22.49

0.39

1.8

3.50

SA-MSTCN\(^{1}\)

3.21

92.2

20.74

3.58

93.6

21.92

0.37

1.4

1.18

SA-MSTCN\(^{2}\)

3.26

92.3

21.00

3.60

93.8

22.21

0.34

1.5

1.21

Mix

Unprocessed

2.08

91.7

14.81

2.08

91.7

14.81

-

-

-

CRN

2.55

93.8

19.29

2.55

93.8

19.29

0.00

0.0

0.00

MSTCN

2.63

93.5

16.40

2.77

94.3

16.82

0.14

0.8

0.42

LSTM-IRM

2.82

94.6

19.26

2.90

95.2

19.90

0.08

0.6

0.64

GCRN

2.86

94.4

20.81

2.85

94.4

20.83

− 0.01

0.0

0.02

GaGNet

2.83

94.0

19.84

2.98

94.9

21.04

0.15

0.9

1.20

Conv-TasNet

2.96

94.6

20.55

2.99

95.0

21.50

0.03

0.4

0.95

DCCRN

3.08

95.0

20.65

3.22

95.7

21.48

0.14

0.7

0.83

DPCRN

3.14

95.2

20.98

3.19

95.6

21.53

0.05

0.4

0.55

SA-MSTCN\(^{1}\)

3.24

95.7

20.98

3.38

96.1

21.45

0.14

0.4

0.47

SA-MSTCN\(^{2}\)

3.26

95.8

21.30

3.41

96.2

21.95

0.15

0.4

0.65