Publication Details
iVector Approach to Phonotactic Language Recognition
Kockmann Marcel, Dipl.-Ing., Ph.D.
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Plchot Oldřich, Ing., Ph.D. (DCGM)
Glembek Ondřej, Ing., Ph.D.
Svendsen Torbjorn
language recognition, subspace modeling, multinomial distribution
We proposed a novel method to extract the iVectors by means of subspace multinomial modelling of the n-gram counts. Using the proposed subspace model, the huge vector of the n-gram counts are represented by the low-dimensional iVector while preserving the discriminative power of the vector.
This paper addresses a novel technique for representation and processing of n-gram counts in phonotactic language recognition (LRE): subspace multinomial modelling represents the vectors of n-gram counts by low dimensional vectors of coordinates in total variability subspace, called iVector. Two techniques for iVector scoring are tested: support vector machines (SVM), and logistic regression (LR). Using standard NIST LRE 2009 task as our evaluation set, the latter scoring approach was shown to outperform phonotactic LRE system based on direct SVM classification of n-gram count vectors. The proposed iVector paradigm also shows comparable results to previously proposed PCA-based phonotactic feature extraction.
@inproceedings{BUT76439,
author="Mehdi Mohammad {Soufifar} and Marcel {Kockmann} and Lukáš {Burget} and Oldřich {Plchot} and Ondřej {Glembek} and Torbjorn {Svendsen}",
title="iVector Approach to Phonotactic Language Recognition",
booktitle="Proceedings of Interspeech 2011",
year="2011",
journal="Proceedings of Interspeech",
volume="2011",
number="8",
pages="2913--2916",
publisher="International Speech Communication Association",
address="Florence",
isbn="978-1-61839-270-1",
issn="1990-9772",
url="http://www.fit.vutbr.cz/research/groups/speech/publi/2011/soufifar_interspeech2011_703.pdf"
}