Speaker Characterization Using TDNN-LSTM Based Speaker Embedding

Chia Ping Chen; Su Yu Zhang; Chih Ting Yeh; Jia Ching Wang; Tenghui Wang; Chien Lin Huang

doi:10.1109/ICASSP.2019.8683185

Speaker Characterization Using TDNN-LSTM Based Speaker Embedding

Chia Ping Chen, Su Yu Zhang, Chih Ting Yeh, Jia Ching Wang, Tenghui Wang, Chien Lin Huang

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

26 Scopus citations

Abstract

In this paper we propose speaker characterization using time delay neural networks and long short-term memory neural networks (TDNN-LSTM) speaker embedding. Three types of front-end feature extraction are investigated to find good features for speaker embedding. Three kinds of data augmentation are used to increase the amount and diversity of the training data. The proposed methods are evaluated with the National Institute of Standards and Technology (NIST) speaker recognition evaluation (SRE) tasks. Experimental results show that the proposed methods achieve a decision cost of 0.400 with the pooled SRE 2018 development set with a single system. In addition, by applying simple average score combination on the outputs of 12 systems, the proposed methods achieve an equal error rate (EER) of 5.56% and a minimum decision cost function of 0.423 with the SRE 2016 evaluation set.

Original language	English
Title of host publication	2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	6211-6215
Number of pages	5
ISBN (Electronic)	9781479981311
DOIs	https://doi.org/10.1109/ICASSP.2019.8683185
State	Published - May 2019
Event	44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Brighton, United Kingdom Duration: 12 May 2019 → 17 May 2019

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume	2019-May
ISSN (Print)	1520-6149

Conference

Conference	44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019
Country/Territory	United Kingdom
City	Brighton
Period	12/05/19 → 17/05/19

Keywords

NIST SRE2018
TDNN-LSTM
speaker embedding

Access to Document

10.1109/ICASSP.2019.8683185

Deep Intelligence Based Spoken Language Processing( II )
Wang, J.-C. (PI)
1/01/19 → 31/12/19
Project: Research

Cite this

Chen, C. P., Zhang, S. Y., Yeh, C. T., Wang, J. C., Wang, T., & Huang, C. L. (2019). Speaker Characterization Using TDNN-LSTM Based Speaker Embedding. In 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings (pp. 6211-6215). Article 8683185 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2019-May). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP.2019.8683185

Chen, Chia Ping ; Zhang, Su Yu ; Yeh, Chih Ting et al. / Speaker Characterization Using TDNN-LSTM Based Speaker Embedding. 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 6211-6215 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{5227400ddf3c459c897bd8f404414593,

title = "Speaker Characterization Using TDNN-LSTM Based Speaker Embedding",

abstract = "In this paper we propose speaker characterization using time delay neural networks and long short-term memory neural networks (TDNN-LSTM) speaker embedding. Three types of front-end feature extraction are investigated to find good features for speaker embedding. Three kinds of data augmentation are used to increase the amount and diversity of the training data. The proposed methods are evaluated with the National Institute of Standards and Technology (NIST) speaker recognition evaluation (SRE) tasks. Experimental results show that the proposed methods achieve a decision cost of 0.400 with the pooled SRE 2018 development set with a single system. In addition, by applying simple average score combination on the outputs of 12 systems, the proposed methods achieve an equal error rate (EER) of 5.56% and a minimum decision cost function of 0.423 with the SRE 2016 evaluation set.",

keywords = "NIST SRE2018, TDNN-LSTM, speaker embedding",

author = "Chen, {Chia Ping} and Zhang, {Su Yu} and Yeh, {Chih Ting} and Wang, {Jia Ching} and Tenghui Wang and Huang, {Chien Lin}",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 ; Conference date: 12-05-2019 Through 17-05-2019",

year = "2019",

month = may,

doi = "10.1109/ICASSP.2019.8683185",

language = "???core.languages.en_GB???",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "6211--6215",

booktitle = "2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings",

}

Chen, CP, Zhang, SY, Yeh, CT, Wang, JC, Wang, T & Huang, CL 2019, Speaker Characterization Using TDNN-LSTM Based Speaker Embedding. in 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings., 8683185, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2019-May, Institute of Electrical and Electronics Engineers Inc., pp. 6211-6215, 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019, Brighton, United Kingdom, 12/05/19. https://doi.org/10.1109/ICASSP.2019.8683185

Speaker Characterization Using TDNN-LSTM Based Speaker Embedding. / Chen, Chia Ping; Zhang, Su Yu; Yeh, Chih Ting et al.
2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. p. 6211-6215 8683185 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2019-May).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Speaker Characterization Using TDNN-LSTM Based Speaker Embedding

AU - Chen, Chia Ping

AU - Zhang, Su Yu

AU - Yeh, Chih Ting

AU - Wang, Jia Ching

AU - Wang, Tenghui

AU - Huang, Chien Lin

PY - 2019/5

Y1 - 2019/5

N2 - In this paper we propose speaker characterization using time delay neural networks and long short-term memory neural networks (TDNN-LSTM) speaker embedding. Three types of front-end feature extraction are investigated to find good features for speaker embedding. Three kinds of data augmentation are used to increase the amount and diversity of the training data. The proposed methods are evaluated with the National Institute of Standards and Technology (NIST) speaker recognition evaluation (SRE) tasks. Experimental results show that the proposed methods achieve a decision cost of 0.400 with the pooled SRE 2018 development set with a single system. In addition, by applying simple average score combination on the outputs of 12 systems, the proposed methods achieve an equal error rate (EER) of 5.56% and a minimum decision cost function of 0.423 with the SRE 2016 evaluation set.

AB - In this paper we propose speaker characterization using time delay neural networks and long short-term memory neural networks (TDNN-LSTM) speaker embedding. Three types of front-end feature extraction are investigated to find good features for speaker embedding. Three kinds of data augmentation are used to increase the amount and diversity of the training data. The proposed methods are evaluated with the National Institute of Standards and Technology (NIST) speaker recognition evaluation (SRE) tasks. Experimental results show that the proposed methods achieve a decision cost of 0.400 with the pooled SRE 2018 development set with a single system. In addition, by applying simple average score combination on the outputs of 12 systems, the proposed methods achieve an equal error rate (EER) of 5.56% and a minimum decision cost function of 0.423 with the SRE 2016 evaluation set.

KW - NIST SRE2018

KW - TDNN-LSTM

KW - speaker embedding

UR - http://www.scopus.com/inward/record.url?scp=85068979447&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2019.8683185

DO - 10.1109/ICASSP.2019.8683185

M3 - 會議論文篇章

AN - SCOPUS:85068979447

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 6211

EP - 6215

BT - 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019

Y2 - 12 May 2019 through 17 May 2019

ER -

Chen CP, Zhang SY, Yeh CT, Wang JC, Wang T, Huang CL. Speaker Characterization Using TDNN-LSTM Based Speaker Embedding. In 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2019. p. 6211-6215. 8683185. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP.2019.8683185

Speaker Characterization Using TDNN-LSTM Based Speaker Embedding

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Projects

Deep Intelligence Based Spoken Language Processing( II )

Cite this