Table 6.
EER obtained on development set (ZTNorm – male and NoNorm – female), comparing classical parameters in a gender-independent setup with a gender-dependent setup in which extra parameters and extended biometric parameters are incorporated [RR → Relative Reduction/[threshold]/(p-value)].
| Parameters | Genre | GSE + VTE setup | Extra parameters | EERM [θM] (p-value) | EERM RR | EERF [θF] (p-value) | EERF RR | HEER [RR] |
|---|---|---|---|---|---|---|---|---|
| Gender-independent configuration (GIC MFCCs + Δ) | M/F | – | – | 10.594% [1.556] | – | 11.693% [−0.009] | – | 11.143% [–] |
| Gender-dependent configuration (GDC MFCCs + Δ + Extra) | M | – | ΔE + F0 + F3 | 9.165% [1.597] (2.02 × 10−5) | 13.5% | 10.183% [8.61%] | ||
| F | – | F0 + F3 | 11.201% [0.016] (2.36 × 10−1) | 6.37% | ||||
| Gender-dependent configuration (GDC MFCCs + Δ + Extra + GSE) | M |
Source-tract sep. Alg Prediction order: 24 Forgetting factor: 0.995 GSE 7-Channel Filter bank 6 MFCC |
ΔE + F0 + F3 | 8.332% [1.619] (1.38 × 10−8) | 21.3% | 9.48% [14.92%] | ||
| F |
Source-tract sep. Alg Prediction order: 36 Forgetting factor: 0.995 GSE 22-Channel Filter bank 4 MFCC |
F0 + F3 | 10.643% [0.010] (4.84 × 10−4) | 8.98% | ||||
| Gender-dependent configuration (GDC MFCCs + Δ + Extra + GSE + VTE) | M | VTE | ΔE + F0 + F3 | 8.496% [1.506] (2.09 × 10−5) | 19.8% | 9.75% [12.50%] | ||
| 14-Channel | ||||||||
| Filter bank | ||||||||
| 2 MFCC | ||||||||
| F | VTE | F0 + F3 | 11.016% [0.023] (2.02 × 10−1) | 5.79% | ||||
| 25-Channel | ||||||||
| Filter bank | ||||||||
| 2 MFCC |
The results highlighted in green, are the ones achieving higher recognition rates.