Table 2 Statistics of different datasets
From: Anchor voiceprint recognition in live streaming via RawNet-SA and gated recurrent unit
| Dataset | VoxCeleb1 | VoxCeleb2 | CN-Celeb |
|---|---|---|---|
| # of POIs | 1251 | 6112 | 1000 |
| # of utterances | 153,516 | 1,128,246 | 130,000 |
| # of hours | 352 | 2442 | 274 |