Publication Details
Migrating i-vectors Between Speaker Recognition Systems Using Regression Neural Networks
Matějka Pavel, Ing., Ph.D. (DCGM)
Plchot Oldřich, Ing., Ph.D. (DCGM)
Pešán Jan, Ing. (DCGM)
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Schwarz Petr, Ing., Ph.D. (DCGM)
speaker recognition, i-vector transformation, Regression Neural Networks, system migration
We have shown that a linear transformation can be used to transform alien i-vectors to the reference i-vectors as the input to the reference PLDA system.
This paper studies the scenario of migrating from one ivector- based speaker recognition system (SRE) to another, i.e. comparing the i-vectors produced by one system with those produced by another system. System migration would typically be motivated by deploying a system with improved recognition accuracy, e.g. because of technological upgrade, or because of the necessity of processing new kind of data, etc. Unfortunately, such migration is very likely to result in the incompatibility between the new and the original i-vectors and, therefore, in the inability of comparing the two. This work studies various topologies of Regression Neural Networks for transforming ivectors from three different systems so that-with slight loss in the accuracy-they are compatible with the reference system. We present the results on the NIST SRE 2010 telephone condition.
@inproceedings{BUT119904,
author="Ondřej {Glembek} and Pavel {Matějka} and Oldřich {Plchot} and Jan {Pešán} and Lukáš {Burget} and Petr {Schwarz}",
title="Migrating i-vectors Between Speaker Recognition Systems Using Regression Neural Networks",
booktitle="Proceedings of Interspeech 2015",
year="2015",
journal="Proceedings of Interspeech",
volume="2015",
number="09",
pages="2327--2331",
publisher="International Speech Communication Association",
address="Dresden",
isbn="978-1-5108-1790-6",
issn="1990-9772",
url="https://www.fit.vut.cz/research/publication/10968/"
}