Estimation of general identifiable linear dynamic models with an application in speech recognition

被引:0
|
作者
Tsontzos, G. [1 ]
Diakoloukas, V. [1 ]
Koniaris, Ch. [1 ]
Digalakis, V. [1 ]
机构
[1] Tech Univ Crete, Dept Elect & Comp Engn, GR-73100 Khania, Greece
来源
2007 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, VOL IV, PTS 1-3 | 2007年
关键词
speech recognition; modeling; identification;
D O I
暂无
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
Although Hidden Markov Models (HMMs) provide a relatively efficient modeling framework for speech recognition, they suffer from several shortcomings which set upper bounds in the performance that can be achieved. Alternatively, linear dynamic models (LDM) can be used to model speech segments. Several implementations of LDM have been proposed in the literature. However, all had a restricted structure to satisfy identifiability constraints. In this paper, we relax all these constraints and use a general, canonical form for a linear state-space system that guarantees identifiability for arbitrary state and observation vector dimensions. For this system, we present a novel, element-wise Maximum Likelihood (ML) estimation method. Classification experiments on the AURORA2 speech database show performance gains compared to HMMs, particularly on highly noisy conditions.
引用
收藏
页码:453 / +
页数:2
相关论文
共 50 条
  • [21] Penalized estimation for non-identifiable models
    Yoshida, Junichiro
    Yoshida, Nakahiro
    ANNALS OF THE INSTITUTE OF STATISTICAL MATHEMATICS, 2024, 76 (05) : 765 - 796
  • [22] RECURSIVE ESTIMATION OF DYNAMIC LINEAR-MODELS
    SNYDER, RD
    JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-METHODOLOGICAL, 1985, 47 (02): : 272 - 276
  • [23] Estimation of variance components in dynamic linear models
    Zacks, S
    Wang, XD
    STATISTICS & PROBABILITY LETTERS, 1999, 41 (03) : 325 - 330
  • [24] Noise robust speech recognition with a switching linear dynamic model
    Droppo, J
    Acero, A
    2004 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, VOL I, PROCEEDINGS: SPEECH PROCESSING, 2004, : 953 - 956
  • [25] Linear trajectory models incorporating preprocessing parameters for speech recognition
    Chengalvarayan, R
    IEEE SIGNAL PROCESSING LETTERS, 1998, 5 (03) : 66 - 68
  • [26] Structured Log Linear Models for Noise Robust Speech Recognition
    Zhang, Shi-Xiong
    Ragni, Anton
    Gales, Mark John Francis
    IEEE SIGNAL PROCESSING LETTERS, 2010, 17 (11) : 945 - 948
  • [27] ON THE EXISTENCE OF IDENTIFIABLE REPARAMETRIZATIONS FOR LINEAR COMPARTMENT MODELS
    Baaijens, Jasmijn A.
    Draisma, Jan
    SIAM JOURNAL ON APPLIED MATHEMATICS, 2016, 76 (04) : 1577 - 1605
  • [28] Convolutional density estimation in hidden Markov models for speech recognition
    Matsoukas, S
    Zavaliagkos, G
    ICASSP '99: 1999 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, PROCEEDINGS VOLS I-VI, 1999, : 113 - 116
  • [29] Convolutional density estimation in hidden Markov models for speech recognition
    Matsoukas, Spyros
    Zavaliagkos, George
    ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, 1999, 1 : 113 - 116
  • [30] ESTIMATION OF PARAMETERS IN DYNAMIC GENERAL LINEAR-MODEL
    HUYBERECHTS, S
    RAIRO-RECHERCHE OPERATIONNELLE-OPERATIONS RESEARCH, 1979, 13 (02): : 143 - 149