Table 14 Performance comparison of RP + MFCC and RP + GFCC bimodal systems over stand-alone nonlinear (RP) and linear (MFCC and GFCC) systems. Note that stand-alone systems over MFCC repeated here for comparison.

From: Recurrence plot embeddings as short segment nonlinear features for multimodal speaker identification using air, bone and throat microphones

Feature

Air–bone

Bone–throat

Air–throat

Air–bone–throat

Val_Acc%

Test_Acc%

Val_Acc%

Test_Acc%

Val_Acc%

Test_Acc%

Val_Acc%

Test_Acc%

RP

97.05 ± 0.19

97.35 ± 0.07

97.90 ± 0.22

97.56 ± 0.1

98.14 ± 0.38

98.21 ± 0.14

99.92 ± 0.2

99.84 ± 0.03

MFCC + RP

97.05 ± 0.02

97.35 ± 0.01

97.91 ± 0.04

97.56 ± 0.24

98.14 ± 0.33

98.21 ± 0.47

99.38± 0.14

99.53 ± 0.05

GFCC + RP

92.56 ± 0.32

91.77 ± 0.87

93.45 ± 0.61

93.54 ± 0.16

89.31 ± 0.72

88.78 ± 0.52

93.66 ± 0.19

93.97 ± 0.38

MFCC + GFCC

97.89 ± 0.30

97.47 ± 0.27

99.09 ± 0.01

98.44 ± 0.06

94.76 ± 0.22

96.11 ± 0.16

98.43 ± 0.11

98.57 ± 0.04

  1. Significant values are in [bold].