Skip to main content
EURASIP Journal on Advances in Signal Processing

Table 2 segSNR improvement comparisons of DDBSE and DD using recursive estimation

From: A priori SNR estimation and noise estimation for speech enhancement

Noise

0 dB

5 dB

10 dB

DD

DDBSE

DD

DDBSE

DD

DDBSE

Babble

2.533

2.722

0.661

0.963

−1.694

−1.417

Car

2.300

3.364

1.158

1.509

−1.15

−0.930

Exhibition

2.586

2.857

0.702

1.006

−1.687

−1.411

Station

2.707

2.973

0.825

1.118

−1.482

−1.216

White

3.187

3.652

1.369

1.783

−0.965

−0.671

Average

2.663

3.114

0.943

1.276

−1.396

−1.129

  1. The numbers in italics indicate the best performance. And all the results are the segSNR improvement compared to that of the untreated noisy speech
Back to article page