Table 2 Datasets used in the experiments. The adaptation set is only used for NIST SRE 2016 and 2018
From: Introducing phonetic information to speaker embedding for speaker verification
Test set | Sampling rate | Phonetic training set | Speaker training set | Adaptation set |
|---|---|---|---|---|
SRE 2010 core-extended & 10s-10s condition 5 | 8kHz | Switchboard-1 Fisher English | Switchboard-2 Phase 1/2/3 Switchboard Cellular Part 1/2 NIST SRE 2004-2008 | - |
SRE 2016 | SRE16 unlabeled data | |||
SRE 2018 | SRE18 unlabeled data | |||
VoxCeleb1 test set | 16kHz | Librispeech | VoxCeleb1 dev set VoxCeleb2 | - |