Analysis of I-vector Length Normalization in Speaker Recognition Systems

被引：0
|
作者：
Garcia-Romero, Daniel [1 ]
Espy-Wilson, Carol Y. [1 ]
机构：
[1] Univ Maryland, Dept Elect & Comp Engn, College Pk, MD 20742 USA
来源：
12TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION 2011 (INTERSPEECH 2011), VOLS 1-5 | 2011年
关键词：
speaker recognition; i-vectors; length normalization;
D O I：
暂无
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
We present a method to boost the performance of probabilistic generative models that work with i-vector representations. The proposed approach deals with the non-Gaussian behavior of i-vectors by performing a simple length normalization. This non-linear transformation allows the use of probabilistic models with Gaussian assumptions that yield equivalent performance to that of more complicated systems based on Heavy-Tailed assumptions. Significant performance improvements are demonstrated on the telephone portion of NIST SEE 2010.
引用
收藏
页码：256 / 259
页数：4
相关论文
共 50 条
  • [31] Full multicondition training for robust i-vector based speaker recognition
    Ribas, Dayana
    Vincent, Emmanuel
    Ramon Calvo, Jose
    [J]. 16TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2015), VOLS 1-5, 2015, : 1057 - 1061
  • [32] DISCRIMINATIVELY RE-TRAINED I-VECTOR EXTRACTOR FOR SPEAKER RECOGNITION
    Novotny, Ondrej
    Plchot, Oldrich
    Glembek, Ondrej
    Burget, Lukas
    Matejka, Pavel
    [J]. 2019 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2019, : 6031 - 6035
  • [33] Nonlinear I-Vector Transformations for PLDA-Based Speaker Recognition
    Cumani, Sandro
    Laface, Pietro
    [J]. IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2017, 25 (04) : 908 - 919
  • [34] Deep Learning Backend for Single and Multisession i-Vector Speaker Recognition
    Ghahabi, Omid
    Hernando, Javier
    [J]. IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2017, 25 (04) : 807 - 817
  • [35] Speaker Recognition Based on i-Vector and Improved Local Preserving Projection
    Wu, Di
    [J]. PROCEEDINGS OF THE 2015 CHINESE INTELLIGENT AUTOMATION CONFERENCE: INTELLIGENT INFORMATION PROCESSING, 2015, 336 : 115 - 121
  • [36] Effect of multicondition training on i-vector PLDA configurations for speaker recognition
    Rajan, Padmanabhan
    Kinnunen, Tomi
    Hautamaki, Ville
    [J]. 14TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2013), VOLS 1-5, 2013, : 3661 - 3664
  • [37] An I-Vector Backend for Speaker Verification
    Kenny, Patrick
    Stafylakis, Themos
    Alam, Jahangir
    Kockmann, Marcel
    [J]. 16TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2015), VOLS 1-5, 2015, : 2307 - 2311
  • [38] A NOISE ROBUST I-VECTOR EXTRACTOR USING VECTOR TAYLOR SERIES FOR SPEAKER RECOGNITION
    Lei, Yun
    Burget, Lukas
    Scheffer, Nicolas
    [J]. 2013 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2013, : 6788 - 6791
  • [39] Improving Short Utterance based I-vector Speaker Recognition using Source and Utterance-Duration Normalization Techniques
    Kanagasundaram, A.
    Dean, D.
    Gonzalez-Dominguez, J.
    Sridharan, S.
    Ramos, D.
    Gonzalez-Rodriguez, J.
    [J]. 14TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2013), VOLS 1-5, 2013, : 2464 - 2468
  • [40] Sparsity Analysis and Compensation for i-Vector Based Speaker Verification
    Li, Wei
    Fu, Tian Fan
    Zhu, Jie
    Chen, Ning
    [J]. SPEECH AND COMPUTER (SPECOM 2015), 2015, 9319 : 381 - 388
12345