Skip to main content

Table 1 Speech denoising SDR scores

From: Blind source separation with optimal transport non-negative matrix factorization

 

OT

OT + OT filter

KL

IS

E

 

k

   

k

   

k

   

k

   

k

   
 

5

10

15

20

5

10

15

20

5

10

15

20

5

10

15

20

5

10

15

20

Cicada

7.7

8.8

8.9

8.4

7.3

8.0

8.6

8.1

7.7

7.9

7.9

7.9

7.7

7.9

7.6

7.7

7.9

8.0

7.9

7.5

 

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

± 0.1

Drums

1.3

3.6

2.7

2.8

1.5

3.8

2.7

2.9

2.0

3.3

2.6

3.1

0.5

0.9

0.6

1.0

1.9

3.4

2.0

2.0

 

± 0.6

± 0.7

± 0.7

± 0.5

± 0.5

± 0.6

± 0.7

± 0.6

± 0.3

± 0.7

± 0.4

± 0.4

± 0.2

± 0.1

± 0.1

± 0.0

± 0.6

± 0.5

± 0.4

± 0.3

Sea

0.0

1.5

3.3

1.8

0.0

1.8

3.3

1.9

1.6

3.4

4.6

4.3

1.6

3.0

3.7

3.5

3.5

4.1

4.4

3.8

 

± 0.9

± 0.7

± 0.5

± 1.1

± 0.8

± 0.6

± 0.5

± 1.0

± 1.3

± 1.0

± 0.8

± 0.7

± 0.8

± 0.6

± 0.6

± 0.5

± 1.1

± 0.9

± 0.9

± 0.6

Subway

2.0

2.8

1.5

2.2

1.8

2.8

1.6

2.3

1.8

2.0

1.9

1.8

2.0

1.4

1.7

2.1

1.5

1.8

1.7

1.7

 

± 1.1

± 0.9

± 0.9

± 1.2

± 1.0

± 1.0

± 0.9

± 1.2

± 1.3

± 1.6

± 0.9

± 0.9

± 0.6

± 0.3

± 0.3

± 0.4

± 1.9

± 1.2

± 1.0

± 0.9

  1. The bold figure in each line indicates the best score for a specific noise