Speaker Identification on VCTK (Mixed)
77AccuracyMFCC GRU
Evaluation results
| Method | Links | |
|---|---|---|
| MFCC GRUInput feature type=MFCC, # layers=1, hidden / layer=128, # parameters (≈, k)=68, Min dilations=1, Max dilations=12017.10 | 77 | |
| Dilated GRUInput feature type=Raw waveform, # layers=8, hidden / layer=50, # parameters (≈, k)=133, Min dilations=8, Max dilations=10242017.10 | 74 | |
| MFCC GRUInput feature type=MFCC, # layers=5, hidden / layer=20, # parameters (≈, k)=16, Min dilations=1, Max dilations=12017.10 | 66 | |
| Fused GRUInput feature type=Raw waveform, # layers=1, hidden / layer=256, # parameters (≈, k)=225, Min dilations=8, Max dilations=82017.10 | 65 | |
| Dilated GRUInput feature type=Raw waveform, # layers=6, hidden / layer=50, # parameters (≈, k)=103, Min dilations=32, Max dilations=10242017.10 | 64 | |
| Fused GRUInput feature type=Raw waveform, # layers=1, hidden / layer=256, # parameters (≈, k)=225, Min dilations=32, Max dilations=322017.10 | 45 |