TY - JOUR
T1 - Autoencoder based domain adaptation for speaker recognition under insufficient channel information
AU - Shon, Suwon
AU - Mun, Seongkyu
AU - Kim, Wooil
AU - Ko, Hanseok
N1 - Funding Information:
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No. 2017R1A2B4012720). This subject is supported by Korea Ministry of Environment (MOE) as “Public Technology Program based on Environmental Policy”.
Publisher Copyright:
Copyright © 2017 ISCA.
Copyright:
Copyright 2017 Elsevier B.V., All rights reserved.
PY - 2017
Y1 - 2017
N2 - In real-life conditions, mismatch between development and test domain degrades speaker recognition performance. To solve the issue, many researchers explored domain adaptation approaches using matched in-domain dataset. However, adaptation would be not effective if the dataset is insufficient to estimate channel variability of the domain. In this paper, we explore the problem of performance degradation under such a situation of insufficient channel information. In order to exploit limited in-domain dataset effectively, we propose an unsupervised domain adaptation approach using Autoencoder based Domain Adaptation (AEDA). The proposed approach combines an autoencoder with a denoising autoencoder to adapt resource-rich development dataset to test domain. The proposed technique is evaluated on the Domain Adaptation Challenge 13 experimental protocols that is widely used in speaker recognition for domain mismatched condition. The results show significant improvements over baselines and results from other prior studies.
AB - In real-life conditions, mismatch between development and test domain degrades speaker recognition performance. To solve the issue, many researchers explored domain adaptation approaches using matched in-domain dataset. However, adaptation would be not effective if the dataset is insufficient to estimate channel variability of the domain. In this paper, we explore the problem of performance degradation under such a situation of insufficient channel information. In order to exploit limited in-domain dataset effectively, we propose an unsupervised domain adaptation approach using Autoencoder based Domain Adaptation (AEDA). The proposed approach combines an autoencoder with a denoising autoencoder to adapt resource-rich development dataset to test domain. The proposed technique is evaluated on the Domain Adaptation Challenge 13 experimental protocols that is widely used in speaker recognition for domain mismatched condition. The results show significant improvements over baselines and results from other prior studies.
KW - Autoencoder
KW - Denoising autoencoder
KW - Domain mismatch
KW - Speaker recognition
KW - Unsupervised domain adaptation
UR - http://www.scopus.com/inward/record.url?scp=85039154266&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85039154266&partnerID=8YFLogxK
U2 - 10.21437/Interspeech.2017-49
DO - 10.21437/Interspeech.2017-49
M3 - Conference article
AN - SCOPUS:85039154266
VL - 2017-August
SP - 1014
EP - 1018
JO - Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH
JF - Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH
SN - 2308-457X
T2 - 18th Annual Conference of the International Speech Communication Association, INTERSPEECH 2017
Y2 - 20 August 2017 through 24 August 2017
ER -