Filters








1,960 Hits in 3.5 sec

Spoken language mismatch in speaker verification: An investigation with NIST-SRE and CRSS Bi-Ling corpora

Abhinav Misra, John H. L. Hansen
2014 2014 IEEE Spoken Language Technology Workshop (SLT)  
Compensation for mismatch between acoustic conditions in automatic speaker recognition has been widely addressed in recent years.  ...  However, performance degradation due to language mismatch has yet to be thoroughly addressed. In this study, we address langauge mismatch for speaker verification.  ...  Using a language detector to detect the language of the test utterance and then choosing an appropriate speaker model trained on that language for scoring [6] .  ... 
doi:10.1109/slt.2014.7078603 dblp:conf/slt/MisraH14 fatcat:bcozgyrfyras5bys2rptrsufg4

TÜRKÇE KONUŞMACI DOĞRULAMADA DİL UYUMSUZLUĞUNUN ETKİSİ

Cemal HANİLÇİ
2017 Uludağ University Journal of The Faculty of Engineering  
In this paper, effect of language mismatch between background data and evaluation data is analyzed for text-independent speaker recognition in particular for Turkish spoken language.  ...  Experiments conducted on a Turkish speech database consisting of 47 male and 26 female speakers reveals that Turkish speaker recognition performance dramatically degrades in case of language mismatch between  ...  The UBM for the speaker verification task is trained using English and Turkish data for the investigation of language mismatch on Turkish speaker verification.  ... 
doi:10.17482/uumfd.309477 fatcat:6f6su3znwbcyrny2anplatnxda

The effect of language factors for robust speaker recognition

Liang Lu, Yuan Dong, Xianyu Zhao, Jiqing Liu, Haila Wang
2009 2009 IEEE International Conference on Acoustics, Speech and Signal Processing  
Based on the conventional joint factor analysis model, we enrolled in the language factors which are mean to capture the language character of each testing and training speech utterance, and compensation  ...  From the results of the NIST speaker recognition evaluation in resent years, speaker recognition systems which are mainly developed based on English training data suffer the language gap problem, namely  ...  Section 2 generally describes joint factor analysis model for speaker verification, and in section 3, we discuss the language factors in JFA in detail.  ... 
doi:10.1109/icassp.2009.4960559 dblp:conf/icassp/LuDZLW09 fatcat:wd5cv3ly3zaptf6wgvsmhqxeoe

Cross-lingual Speaker Verification with Deep Feature Learning [article]

Lantian Li and Dong Wang and Askar Rozi and Thomas Fang Zheng
2017 arXiv   pre-print
Existing speaker verification (SV) systems often suffer from performance degradation if there is any language mismatch between model training, speaker enrollment, and test.  ...  In this paper, we investigate the robustness of the feature-based SV system in situations with language mismatch.  ...  Conclusions This paper investigated the feature-based speaker verification approach in situations with language mismatch between training, enrollment and test.  ... 
arXiv:1706.07861v1 fatcat:pqfc227j7redfmnzdduded2w3u

Cross-lingual Text-independent Speaker Verification Using Unsupervised Adversarial Discriminative Domain Adaptation

Wei Xia, Jing Huang, John H.L. Hansen
2019 ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
Speaker verification systems often degrade significantly when there is a language mismatch between training and testing data.  ...  to the shift in language.  ...  Language mismatch falls into two scenarios that include (i) the speaker verification system is trained on one language, but the enrollment and test data for speakers are in a second language, and (ii)  ... 
doi:10.1109/icassp.2019.8682259 dblp:conf/icassp/XiaHH19 fatcat:ulwrq5klbbad7dfbzdtg3b6sga

Domain-invariant I-vector Feature Extraction for PLDA Speaker Verification

Md Hafizur Rahman, Ivan Himawan, David Dean, Clinton Fookes, Sridha Sridharan
2018 Odyssey 2018 The Speaker and Language Recognition Workshop  
In this paper, we introduce a domain-invariant i-vector extraction (DI-IVEC) approach to extract domain mismatch compensated out-domain i-vectors using limited in-domain (target) data for adaptation.  ...  The proposed method provides at least 17.3% improvement in EER over an out-domain-only trained baseline when speaker labels are absent and a 27.2% improvement in EER when speaker labels are known.  ...  Also, an additional domain mismatch can be compensated in the i-vector subspace when DI-IVEC is used in combination with DICN, thus improving the overall speaker verification performance.  ... 
doi:10.21437/odyssey.2018-22 dblp:conf/odyssey/RahmanHDFS18 fatcat:sl3o3lkasfa47jyknijtoqvjya

Optimization of Telephone Handset Distortion in Speaker Recognition by Discriminative Feature Design

Kadiri Kamoru Oluwatoyin
2014 IOSR Journal of Electronics and Communication Engineering  
Effect of handset distortion was done to maximize speaker recognition performance specifically in the setting of telephone handset mismatch between training and testing as results on the 1998 NIST Speaker  ...  This technique used mel-cepstral features, log spectrum and prosody based features with a nonlinear artificial neural network in designing speaker recognition features that minimize telephone handset distortion  ...  A score-based handset and channel compensation method for speaker recognition systems called HNORM was presented in Reynolds, 1997b .  ... 
doi:10.9790/2834-09523136 fatcat:4xymka7fzfhjhb5xg7vqz4fv5i

An overview of robustness related issues in speaker recognition

Thomas Fang Zheng, Qin Jin, Lantian Li, Jun Wang, Fanhu Bie
2014 Signal and Information Processing Association Annual Summit and Conference (APSIPA), 2014 Asia-Pacific  
For each category, we then describe the current hot topics, existing technologies, and potential research focuses in the future. I. 978-616-361-823-8  ...  We first categorize the robustness issues into three categories, including environment-related, speaker-related and application-oriented issues.  ...  Research dealing with the channel mismatch of speaker verification tasks can be categorized into three directions: feature transformation [7, 19, 20] , model compensation [21, 22] and score normalization  ... 
doi:10.1109/apsipa.2014.7041826 dblp:conf/apsipa/ZhengJLWB14 fatcat:wjnsvr7eavambpucbcxqsftt6a

Cross-lingual speaker verification based on linear transform

Rozi Askar, Dong Wang, Fanhu Bie, Jun Wang, Thomas Fang Zheng
2015 2015 IEEE China Summit and International Conference on Signal and Information Processing (ChinaSIP)  
Speaker verification suffers from serious performance degradation if the enrollment and test speech are in different languages.  ...  This paper proposes a linear transform approach which projects speech signals from its own language to another language so that the language mismatch between enrollment and test can be mitigated.  ...  This language factor is inferred and compensated for during enrollment and test.  ... 
doi:10.1109/chinasip.2015.7230457 dblp:conf/chinasip/AskarWBWZ15 fatcat:nkjh2x2xkfa5lgnt7awenr5naa

Multi-variability speech database for robust speaker recognition

B C Haris, G Pradhan, A Misra, S Shukla, R Sinha, S R M Prasanna
2011 2011 National Conference on Communications (NCC)  
The collected database is evaluated using adapted Gaussian mixture model based speaker verification system following the NIST 2003 speaker recognition evaluation protocol and gives comparable performance  ...  in performance compared to the matched case whereas for language mismatch case the degradation is found to be relatively smaller.  ...  on their modeling or compensation.  ... 
doi:10.1109/ncc.2011.5734775 fatcat:c7mqs3msafebhngoioq6jstz5m

Eigenchannel Compensation and Symmetric Score for Robust Text-Independent Speaker Verification

Yuan Dong, Jian Zhao, Liang Lu, Jiqing Lui, Xianyu Zhao, Haila Wang
2008 2008 6th International Symposium on Chinese Spoken Language Processing  
The negative effect of the session variability has become more and more severe for the performance of the speaker verification system.  ...  This paper discusses the eigenchannel compensation and investigates the symmetric scoring method to diminish the session variability and further enhance the performance.  ...  EIGENCHANNEL COMPENSATION In eigenchannel filtering for text independent speaker verification, the GMM mean vectors for the incoming utterance can be factored in two parts.  ... 
doi:10.1109/chinsl.2008.ecp.92 dblp:conf/iscslp/DongZLLZW08 fatcat:qr7jch5dzfaflfzo7rad45523e

Support Vector Machine Regression for Robust Speaker Verification in Mismatching and Forensic Conditions [chapter]

Ismael Mateos-Garcia, Daniel Ramos, Ignacio Lopez-Moreno, Joaquin Gonzalez-Rodriguez
2009 Lecture Notes in Computer Science  
In this paper we propose the use of Support Vector Machine Regression (SVR) for robust speaker verification in two scenarios: i) strong mismatch in speech conditions and ii) forensic environment.  ...  For the mismatching condition scenario, we use the NIST SRE 2008 core task as a highly variable environment, but with a mostly representative background set coming from past NIST evaluations.  ...  Jose Juan Lucena and people from the Acoustics and Image Processing Department from Guardia Civil for their important effort in collecting data for forensic purposes.  ... 
doi:10.1007/978-3-642-01793-3_50 fatcat:tq3yicnaubac7fwbayony5mgjy

Investigating Deep Neural Networks for Speaker Diarization in the DIHARD Challenge

Ivan Himawan, Md Hafizur Rahman, Sridha Sridharan, Clinton Fookes, Ahilan Kanagasundaram
2018 2018 IEEE Spoken Language Technology Workshop (SLT)  
to compensate the mismatch in the embedding subspace.  ...  In the speaker recognition task, domain adaptation works very well provided long in-domain utterance for the domain mismatch compensation training.  ... 
doi:10.1109/slt.2018.8639630 dblp:conf/slt/HimawanRSFK18 fatcat:5hecj7uukrdirfctzygfmbthbu

An Analysis of Transfer Learning for Domain Mismatched Text-independent Speaker Verification

Chunlei Zhang, Shivesh Ranjan, John Hansen
2018 Odyssey 2018 The Speaker and Language Recognition Workshop  
In this paper, we present transfer learning for deep neural network based text-independent speaker verification, in the presence of a severe mismatch between the enrollment and the test data.  ...  Given a pre-trained speaker embedding network developed with out-of-domain data, we explore and analyze how this pre-trained model can benefit for the in-domain speaker verification task.  ...  model is essential to improve the performance for domainmismatched speaker verification.  ... 
doi:10.21437/odyssey.2018-26 dblp:conf/odyssey/ZhangRH18 fatcat:jo7mzr6qkfejjb6b5qmq2sdukq

A robust speaker recognition system combining factor analysis techniques

Shaghayegh Reza, Tahereh Emami Azadi, Jahanshah Kabudian, Yaser Shekofteh
2014 2014 21th Iranian Conference on Biomedical Engineering (ICBME)  
I-Vector has been proposed and used in the field of language recognition and speaker verification.  ...  JFA is a powerful tool in speaker verification task and models the speaker inter-variabilities and compensate channel and session variabilities.  ... 
doi:10.1109/icbme.2014.7043948 fatcat:mujdmdeu3ja2dm6dhribcoijuq
« Previous Showing results 1 — 15 out of 1,960 results