Skip to main content

Table 1 SIR of hybrid algorithm for the experiment performed in the simulated room environment shown in Figure 5

From: A hybrid algorithm for blind source separation of a convolutive mixture of three speech sources

 

Time (s)

 

0 to 4

4 to 8

8 to 12

12 to 16

16 to 20

20 to 24

24 to 28

28 to 32

32 to 36

36 to 40

40 to 44

SIR (dB)

Input (stream 1)

4.7

0.8

−9.8

4.6

−1.4

0.8

3.2

−4.4

−1.3

−18.3

−3.1

CFDD

5.0

1.8

−4.2

15.8

0.2

4.8

10.4

5.4

12.2

−12.5

7.6

Hybrid

7.5

4.2

−0.7

23.1

10.4

7.7

14.4

9.2

12.7

−9.3

11.7

Input (stream 2)

4.1

−4.1

0.4

−5.4

−5.6

−3.1

−10.3

−8.8

−6.5

0.5

−6.4

CFDD

7.0

−1.9

5.4

0.6

0.5

1.1

2.1

6.0

6.0

11.2

4.0

Hybrid

11.1

2.4

14.7

5.3

10.7

8.4

7.1

10.2

15.9

15.4

8.2

Input (stream 3)

−5.2

−1.8

−5.1

−2.2

−0.3

1.9

−21.4

7.5

−1.9

3.4

0.7

CFDD

−3.2

1.4

3.3

5.7

3.6

8.5

−12.8

21.3

12.0

13.5

14.2

Hybrid

1.0

6.3

10.4

9.8

10.0

13.3

−10.2

25.6

14.5

16.3

15.5

SAR

Average SAR hybrid

32.72

31.24

27.92

26.16

21.13

19.28

18.12

15.71

15.9

16.87

18.24

  1. The whole speech length has been divided into 4-s segments, and SIR and SAR are calculated on 20-ms syllable within that. The room reverberation value is taken to be 0.4 with a room impulse response length of 50 ms.