Table 6 Ablation experiment.

From: TDNN achitecture with efficient channel attention and improved residual blocks for accurate speaker recognition

ECAM

DSConv

PRS

SSS

TS

EER

minDCF

ACC(%)

Params(M)

     

0.179

0.799

89.1

6.9

\(\checkmark\)

    

0.168

0.752

91.9

6.2

 

\(\checkmark\)

   

0.178

0.797

89.8

6.7

  

\(\checkmark\)

  

0.173

0.791

90.2

6.8

   

\(\checkmark\)

 

0.167

0.750

92.5

14.9

\(\checkmark\)

  

\(\checkmark\)

 

0.160

0.741

94.1

14.2

 

\(\checkmark\)

\(\checkmark\)

  

0.171

0.783

91.0

6.5

\(\checkmark\)

 

\(\checkmark\)

\(\checkmark\)

 

0.158

0.731

94.7

14.1

\(\checkmark\)

\(\checkmark\)

\(\checkmark\)

\(\checkmark\)

 

0.156

0.728

95.1

-

\(\checkmark\)

\(\checkmark\)

\(\checkmark\)

\(\checkmark\)

\(\checkmark\)

0.154

0.724

95.7

13.9