Lexicon-Based Local Representation for Text-Dependent Speaker Verification

Hanxu YOU; Wei LI; Lianqiang LI; Jie ZHU

doi:10.1587/transinf.2016EDL8182

Lexicon-Based Local Representation for Text-Dependent Speaker Verification

Hanxu YOU, Wei LI, Lianqiang LI, Jie ZHU

Full Text Views

0

Share
Cite this

Summary :

A text-dependent i-vector extraction scheme and a lexicon-based binary vector (L-vector) representation are proposed to improve the performance of text-dependent speaker verification. I-vector and L-vector are used to represent the utterances for enrollment and test. An improved cosine distance kernel is constructed by combining i-vector and L-vector together and is used to distinguish both speaker identity and lexical (or text) diversity with back-end support vector machine (SVM). Experiments are conducted on RSR 2015 Corpus part 1 and part 2, the results indicate that at most 30% improvement can be obtained compared with traditional i-vector baseline.

Publication: IEICE TRANSACTIONS on Information Vol.E100-D No.3 pp.587-589

Publication Date: 2017/03/01

Publicized: 2016/12/05

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2016EDL8182

Type of Manuscript: LETTER

Category: Speech and Hearing

Authors

Hanxu YOU
  Shanghai Jiao Tong University (SJTU)
Wei LI
  Shanghai Jiao Tong University (SJTU)
Lianqiang LI
  Shanghai Jiao Tong University (SJTU)
Jie ZHU
  Shanghai Jiao Tong University (SJTU)

Keyword

i-vector, L-vector, text-dependent speaker verification, cosine distance kernel

Cite this

Copy

Hanxu YOU, Wei LI, Lianqiang LI, Jie ZHU, "Lexicon-Based Local Representation for Text-Dependent Speaker Verification" in IEICE TRANSACTIONS on Information, vol. E100-D, no. 3, pp. 587-589, March 2017, doi: 10.1587/transinf.2016EDL8182.
Abstract: A text-dependent i-vector extraction scheme and a lexicon-based binary vector (L-vector) representation are proposed to improve the performance of text-dependent speaker verification. I-vector and L-vector are used to represent the utterances for enrollment and test. An improved cosine distance kernel is constructed by combining i-vector and L-vector together and is used to distinguish both speaker identity and lexical (or text) diversity with back-end support vector machine (SVM). Experiments are conducted on RSR 2015 Corpus part 1 and part 2, the results indicate that at most 30% improvement can be obtained compared with traditional i-vector baseline.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2016EDL8182/_p

Copy

@ARTICLE{e100-d_3_587,
author={Hanxu YOU, Wei LI, Lianqiang LI, Jie ZHU, },
journal={IEICE TRANSACTIONS on Information},
title={Lexicon-Based Local Representation for Text-Dependent Speaker Verification},
year={2017},
volume={E100-D},
number={3},
pages={587-589},
abstract={A text-dependent i-vector extraction scheme and a lexicon-based binary vector (L-vector) representation are proposed to improve the performance of text-dependent speaker verification. I-vector and L-vector are used to represent the utterances for enrollment and test. An improved cosine distance kernel is constructed by combining i-vector and L-vector together and is used to distinguish both speaker identity and lexical (or text) diversity with back-end support vector machine (SVM). Experiments are conducted on RSR 2015 Corpus part 1 and part 2, the results indicate that at most 30% improvement can be obtained compared with traditional i-vector baseline.},
keywords={},
doi={10.1587/transinf.2016EDL8182},
ISSN={1745-1361},
month={March},}

Copy

TY - JOUR
TI - Lexicon-Based Local Representation for Text-Dependent Speaker Verification
T2 - IEICE TRANSACTIONS on Information
SP - 587
EP - 589
AU - Hanxu YOU
AU - Wei LI
AU - Lianqiang LI
AU - Jie ZHU
PY - 2017
DO - 10.1587/transinf.2016EDL8182
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E100-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2017
AB - A text-dependent i-vector extraction scheme and a lexicon-based binary vector (L-vector) representation are proposed to improve the performance of text-dependent speaker verification. I-vector and L-vector are used to represent the utterances for enrollment and test. An improved cosine distance kernel is constructed by combining i-vector and L-vector together and is used to distinguish both speaker identity and lexical (or text) diversity with back-end support vector machine (SVM). Experiments are conducted on RSR 2015 Corpus part 1 and part 2, the results indicate that at most 30% improvement can be obtained compared with traditional i-vector baseline.
ER -