Learning Deep Embedding with Acoustic and Phoneme Features for Speaker Recognition in FM Broadcasting

Table 3

TI-SV results of the hybrid DNNs in FM broadcasting.

Method	Loss	Optimization	EER (%)
ours_no_PFE	AAM-Softmax	—	14.29
ours_PFI_1	AAM-Softmax	—	10.71
ours_no_PFE	AAM-Softmax	Fine-tuning	8.93
ours_PFI_1	AAM-Softmax	Fine-tuning	7.14

The use of “bold” is to emphasize the experimental result (7.14%).