English

Attentive Statistics Pooling for Deep Speaker Embedding

Audio and Speech Processing 2019-02-27 v2 Sound

Abstract

This paper proposes attentive statistics pooling for deep speaker embedding in text-independent speaker verification. In conventional speaker embedding, frame-level features are averaged over all the frames of a single utterance to form an utterance-level feature. Our method utilizes an attention mechanism to give different weights to different frames and generates not only weighted means but also weighted standard deviations. In this way, it can capture long-term variations in speaker characteristics more effectively. An evaluation on the NIST SRE 2012 and the VoxCeleb data sets shows that it reduces equal error rates (EERs) from the conventional method by 7.5% and 8.1%, respectively.

Keywords

Cite

@article{arxiv.1803.10963,
  title  = {Attentive Statistics Pooling for Deep Speaker Embedding},
  author = {Koji Okabe and Takafumi Koshinaka and Koichi Shinoda},
  journal= {arXiv preprint arXiv:1803.10963},
  year   = {2019}
}

Comments

Proc. Interspeech 2018, pp2252--2256. arXiv admin note: text overlap with arXiv:1809.09311

R2 v1 2026-06-23T01:08:34.664Z