Skip to main content
EURASIP Journal on Advances in Signal Processing

Table 7 Breakdown of speech recognition errors (%)

From: Dereverberation and denoising based on generalized spectral subtraction by multi-channel LMS algorithm using a small-scale microphone array

 

CMN only

Power SS

GSS (proposed)

  

w/o MFT

MFT

w/o MFT

MFT

Sub

40.61

30.48

29.37

27.39

27.42

Del

13.82

9.27

9.26

8.99

8.06

Ins

3.67

4.44

4.10

3.47

3.67

Back to article page