Latent Dirichlet Allocation is a powerful topic model used heavily in natural language processing, image processing and biomedical signal processing fields to discover hidden structures behind observed data. In this work, we have adopted a variant of LDA for continuous descriptor vectors and use this model as a front-end for speaker verification similar to popular i-vector front-end. We have proposed an efficient hierarchical acoustic vocabulary creation method and presented a speaker verification system using latent topic probability features obtained using LDA front-end. We analysed the performance of the LDA front-end for various vocabulary and topic sizes, and obtained encouraging results on NIST SRE corpora. The proposed system is shown to improve the performance of an ivector-PLDA baseline system when tested on NIST SRE12 corpora.
Cite as: Isik, Y.Z., Erdogan, H., Sarikaya, R. (2014) A Latent Dirichlet Allocation Based Front-End for Speaker Verification. Proc. The Speaker and Language Recognition Workshop (Odyssey 2014), 131-136, doi: 10.21437/Odyssey.2014-12
@inproceedings{isik14_odyssey,
author={Yusuf Ziya Isik and Hakan Erdogan and Ruhi Sarikaya},
title={{A Latent Dirichlet Allocation Based Front-End for Speaker Verification}},
year=2014,
booktitle={Proc. The Speaker and Language Recognition Workshop (Odyssey 2014)},
pages={131--136},
doi={10.21437/Odyssey.2014-12}
}