Publication Details
Fast variational Bayes for heavy-tailed PLDA applied to i-vectors and x-vectors
Brummer Johan Nikolaas Langenhoven, Dr.
García-Romero Daniel
SNYDER, D.
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
peaker recognition, variational Bayes, heavytailed PLDA
The standard state-of-the-art backend for text-independent speaker recognizers that use i-vectors or x-vectors, is Gaussian PLDA (G-PLDA), assisted by a Gaussianization step involving length normalization. G-PLDA can be trained with both generative or discriminative methods. It has long been known that heavy-tailed PLDA (HT-PLDA), applied without length normalization, gives similar accuracy, but at considerable extra computational cost. We have recently introduced a fast scoring algorithm for a discriminatively trained HT-PLDA backend. This paper extends that work by introducing a fast, variational Bayes, generative training algorithm. We compare old and new backends, with and without length-normalization, with i-vectors and x-vectors, on SRE10, SRE16 and SITW.
@inproceedings{BUT155098,
author="SILNOVA, A. and BRUMMER, J. and GARCÍA-ROMERO, D. and SNYDER, D. and BURGET, L.",
title="Fast variational Bayes for heavy-tailed PLDA applied to i-vectors and x-vectors",
booktitle="Proceedings of Interspeech 2018",
year="2018",
journal="Proceedings of Interspeech",
volume="2018",
number="9",
pages="72--76",
publisher="International Speech Communication Association",
address="Hyderabad",
doi="10.21437/Interspeech.2018-2128",
issn="1990-9772",
url="https://www.isca-speech.org/archive/Interspeech_2018/abstracts/2128.html"
}