DNN and i-vector combined method for speaker recognition on multi-variability environments

被引：3
|
作者：
Reyes-Diaz, Flavio J. [1 ]
Hernandez-Sierra, Gabriel [1 ]
de Lara, Jose R. Calvo [1 ]
机构：
[1] Adv Technol Applicat Ctr CENATAV, 7a A 21406 E-214 & 216, Havana 12200, Cuba
来源：
INTERNATIONAL JOURNAL OF SPEECH TECHNOLOGY | 2021年 / 24卷 / 02期
关键词：
Multi-variability compensation; Bottleneck features; Speaker verification; Short utterances; Additive noise; Reverberation; Deep neural network; VERIFICATION;
D O I：
10.1007/s10772-021-09796-1
中图分类号：
TM [电工技术]; TN [电子技术、通信技术];
学科分类号：
0808 ; 0809 ;
摘要：
The article deals with the compensation of variability in Automatic Speaker Verification systems in scenarios where the variability conditions due to utterance duration, reverberation and environmental noise are simultaneously present. We introduce a new representation of the speaker's discriminative information, based on the use of a deep neural network trained discriminatively for speaker classification and i-vector representation. The proposed representation allows us to increase the verification performance by reducing the error between 2.5 and 7.9 % for all variability conditions compared to baseline systems. We also analyze the speaker verification system robustness based on interquartile range, obtaining a 1.19 times improvement compared to baselines evaluated.
引用
收藏
页码：409 / 418
页数：10
相关论文
共 50 条
  • [1] DNN and i-vector combined method for speaker recognition on multi-variability environments
    Flavio J. Reyes-Díaz
    Gabriel Hernández-Sierra
    José R. Calvo de Lara
    [J]. International Journal of Speech Technology, 2021, 24 : 409 - 418
  • [2] IMPROVING DNN SPEAKER INDEPENDENCE WITH I-VECTOR INPUTS
    Senior, Andrew
    Lopez-Moreno, Ignacio
    [J]. 2014 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2014,
  • [3] END-TO-END DNN BASED SPEAKER RECOGNITION INSPIRED BY I-VECTOR AND PLDA
    Rohdin, Johan
    Silnova, Anna
    Diez, Mireia
    Plchot, Oldrich
    Matejka, Pavel
    Burget, Lukas
    [J]. 2018 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2018, : 4874 - 4878
  • [4] On the Complementary Role of DNN Multi-Level Enhancement for Noisy Robust Speaker Recognition in an I-Vector Framework
    Zhang, Xingyu
    Zou, Xia
    Sun, Meng
    Wu, Penglong
    Wang, Yimin
    He, Jun
    [J]. IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES, 2020, E103A (01) : 356 - 360
  • [5] I-vector Based Speaker Gender Recognition
    Wang, Minghe
    Chen, Ying
    Tang, Zhenmin
    Zhang, Erhua
    [J]. 2015 IEEE ADVANCED INFORMATION TECHNOLOGY, ELECTRONIC AND AUTOMATION CONTROL CONFERENCE (IAEAC), 2015, : 729 - 732
  • [6] I-Vector DNN Scoring and Calibration for Noise Robust Speaker Verification
    Tan, Zhili
    Mak, Man-Wai
    [J]. 18TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2017), VOLS 1-6: SITUATED INTERACTION, 2017, : 1562 - 1566
  • [7] i-vector Based Speaker Recognition on Short Utterances
    Kanagasundaram, Ahilan
    Vogt, Robbie
    Dean, David
    Sridharan, Sridha
    Mason, Michael
    [J]. 12TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION 2011 (INTERSPEECH 2011), VOLS 1-5, 2011, : 2352 - +
  • [8] I-Vector Speaker and Language Recognition System on Android
    Vazquez-Machado, Christian
    Colon-Hernandez, Pedro
    Torres-Carrasquillo, Pedro A.
    [J]. 2016 IEEE HIGH PERFORMANCE EXTREME COMPUTING CONFERENCE (HPEC), 2016,
  • [9] Generalizing I-Vector Estimation for Rapid Speaker Recognition
    Xu, Longting
    Lee, Kong Aik
    Li, Haizhou
    Yang, Zhen
    [J]. IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2018, 26 (04) : 749 - 759
  • [10] AN IMPROVED UNCERTAINTY PROPAGATION METHOD FOR ROBUST I-VECTOR BASED SPEAKER RECOGNITION
    Ribas, Dayana
    Vincent, Emmanuel
    [J]. 2019 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2019, : 6331 - 6335
12345