New Background Speaker Models and Experiments on the ANDOSL Speech Corpus

Dat Tran; Dharmendra Sharma

doi:10.1007/978-3-540-30133-2_65

New Background Speaker Models and Experiments on the ANDOSL Speech Corpus

Dat Tran, Dharmendra Sharma

Research output: A Conference proceeding or a Chapter in Book › Conference contribution › peer-review

4 Citations (Scopus)

Abstract

We present a new background speaker modeling method for speaker verification. This method does not require speech databases of other speakers to build background speaker models. Background speaker models can be built directly from the same claimed speaker’s database and have smaller numbers of Gaussian mixtures compared to the claimed speaker model. Experiments performed on the Australian ANDOSL speech database showed better results for speaker verification using the 32-mixture claimed speaker model and 16-mixture background model compared to current background model set methods using five closest and five same-group background models

Original language	English
Title of host publication	International Conference on Knowledge-Based and Intelligent Information and Engineering Systems KES 2004
Editors	M.G Negoita, R.J Howlett, L.C Jain
Place of Publication	Germany
Publisher	Springer
Pages	498-503
Number of pages	2
ISBN (Print)	9783540232063
DOIs	https://doi.org/10.1007/978-3-540-30133-2_65
Publication status	Published - 2004
Event	Knowledge-Based Intelligent Information and Engineering Systems - Wellington, New Zealand Duration: 22 Sept 2004 → 24 Sept 2004

Conference

Conference	Knowledge-Based Intelligent Information and Engineering Systems
Country/Territory	New Zealand
City	Wellington
Period	22/09/04 → 24/09/04

Access to Document

10.1007/978-3-540-30133-2_65

Cite this

@inproceedings{aae17cbc6e70495f9a57602e0c3bcb4d,

title = "New Background Speaker Models and Experiments on the ANDOSL Speech Corpus",

abstract = "We present a new background speaker modeling method for speaker verification. This method does not require speech databases of other speakers to build background speaker models. Background speaker models can be built directly from the same claimed speaker{\textquoteright}s database and have smaller numbers of Gaussian mixtures compared to the claimed speaker model. Experiments performed on the Australian ANDOSL speech database showed better results for speaker verification using the 32-mixture claimed speaker model and 16-mixture background model compared to current background model set methods using five closest and five same-group background models",

author = "Dat Tran and Dharmendra Sharma",

year = "2004",

doi = "10.1007/978-3-540-30133-2_65",

language = "English",

isbn = "9783540232063",

pages = "498--503",

editor = "M.G Negoita and R.J Howlett and L.C Jain",

booktitle = "International Conference on Knowledge-Based and Intelligent Information and Engineering Systems KES 2004",

publisher = "Springer",

address = "Netherlands",

note = "Knowledge-Based Intelligent Information and Engineering Systems ; Conference date: 22-09-2004 Through 24-09-2004",

}

Tran, D & Sharma, D 2004, New Background Speaker Models and Experiments on the ANDOSL Speech Corpus. in MG Negoita, RJ Howlett & LC Jain (eds), International Conference on Knowledge-Based and Intelligent Information and Engineering Systems KES 2004. Springer, Germany, pp. 498-503, Knowledge-Based Intelligent Information and Engineering Systems, Wellington, New Zealand, 22/09/04. https://doi.org/10.1007/978-3-540-30133-2_65

New Background Speaker Models and Experiments on the ANDOSL Speech Corpus. / Tran, Dat ; Sharma, Dharmendra.
International Conference on Knowledge-Based and Intelligent Information and Engineering Systems KES 2004. ed. / M.G Negoita; R.J Howlett; L.C Jain. Germany: Springer, 2004. p. 498-503.

Research output: A Conference proceeding or a Chapter in Book › Conference contribution › peer-review

TY - GEN

T1 - New Background Speaker Models and Experiments on the ANDOSL Speech Corpus

AU - Tran, Dat

AU - Sharma, Dharmendra

PY - 2004

Y1 - 2004

N2 - We present a new background speaker modeling method for speaker verification. This method does not require speech databases of other speakers to build background speaker models. Background speaker models can be built directly from the same claimed speaker’s database and have smaller numbers of Gaussian mixtures compared to the claimed speaker model. Experiments performed on the Australian ANDOSL speech database showed better results for speaker verification using the 32-mixture claimed speaker model and 16-mixture background model compared to current background model set methods using five closest and five same-group background models

AB - We present a new background speaker modeling method for speaker verification. This method does not require speech databases of other speakers to build background speaker models. Background speaker models can be built directly from the same claimed speaker’s database and have smaller numbers of Gaussian mixtures compared to the claimed speaker model. Experiments performed on the Australian ANDOSL speech database showed better results for speaker verification using the 32-mixture claimed speaker model and 16-mixture background model compared to current background model set methods using five closest and five same-group background models

U2 - 10.1007/978-3-540-30133-2_65

DO - 10.1007/978-3-540-30133-2_65

M3 - Conference contribution

SN - 9783540232063

SP - 498

EP - 503

BT - International Conference on Knowledge-Based and Intelligent Information and Engineering Systems KES 2004

A2 - Negoita, M.G

A2 - Howlett, R.J

A2 - Jain, L.C

PB - Springer

CY - Germany

T2 - Knowledge-Based Intelligent Information and Engineering Systems

Y2 - 22 September 2004 through 24 September 2004

ER -

New Background Speaker Models and Experiments on the ANDOSL Speech Corpus

Abstract

Conference

Access to Document

Fingerprint

Cite this