An extension of the state-observation dependency in Partly Hidden Markov Models and its application to continuous speech recognition

    Research output: Contribution to journalArticle

    1 Citation (Scopus)

    Abstract

    We extend the state-observation dependencies in a Partly Hidden Markov Model (PHMM) and apply this model to continuous speech recognition. In a PHMM the observations and state transitions are dependent on a series of hidden and observable states. In the standard formulation of a PHMM, the observations and state transitions are conditioned on the same hidden state and observable state variables. Here we also condition the observations and state transitions on the same hidden states but condition the observations and state transitions on different observation states, respectively. This simple improvement to the model gives it significant flexibility allowing it to model stochastic processes more precisely. In addition, by integrating the PHMM containing this extended state-observation dependency with a standard HMM we can construct a stochastic model that we call a Smoothed Partly Hidden Markov Model (SPHMM). Results of continuous speech recognition on a newspaper read-speech have shown reductions of 10 and 24% in the error rate using the PHMM and SPHMM, respectively, compared to a standard HMM thereby displaying the effectiveness of the proposed models.

    Original languageEnglish
    Pages (from-to)31-39
    Number of pages9
    JournalSystems and Computers in Japan
    Volume36
    Issue number8
    DOIs
    Publication statusPublished - 2005 Jul

    Fingerprint

    Continuous speech recognition
    Hidden Markov models
    Speech Recognition
    Markov Model
    State Transition
    Stochastic models
    Observation
    Random processes
    Model
    Stochastic Model
    Error Rate
    Stochastic Processes
    Flexibility
    Series
    Formulation
    Dependent

    Keywords

    • Acoustic models
    • Continuous speech recognition
    • HMM
    • PHMM
    • SPHMM

    ASJC Scopus subject areas

    • Hardware and Architecture
    • Information Systems
    • Theoretical Computer Science
    • Computational Theory and Mathematics

    Cite this

    @article{feebb99e3204432fb163bbaaca3bf580,
    title = "An extension of the state-observation dependency in Partly Hidden Markov Models and its application to continuous speech recognition",
    abstract = "We extend the state-observation dependencies in a Partly Hidden Markov Model (PHMM) and apply this model to continuous speech recognition. In a PHMM the observations and state transitions are dependent on a series of hidden and observable states. In the standard formulation of a PHMM, the observations and state transitions are conditioned on the same hidden state and observable state variables. Here we also condition the observations and state transitions on the same hidden states but condition the observations and state transitions on different observation states, respectively. This simple improvement to the model gives it significant flexibility allowing it to model stochastic processes more precisely. In addition, by integrating the PHMM containing this extended state-observation dependency with a standard HMM we can construct a stochastic model that we call a Smoothed Partly Hidden Markov Model (SPHMM). Results of continuous speech recognition on a newspaper read-speech have shown reductions of 10 and 24{\%} in the error rate using the PHMM and SPHMM, respectively, compared to a standard HMM thereby displaying the effectiveness of the proposed models.",
    keywords = "Acoustic models, Continuous speech recognition, HMM, PHMM, SPHMM",
    author = "Tetsuji Ogawa and Tetsunori Kobayashi",
    year = "2005",
    month = "7",
    doi = "10.1002/scj.20315",
    language = "English",
    volume = "36",
    pages = "31--39",
    journal = "Systems and Computers in Japan",
    issn = "0882-1666",
    publisher = "John Wiley and Sons Inc.",
    number = "8",

    }

    TY - JOUR

    T1 - An extension of the state-observation dependency in Partly Hidden Markov Models and its application to continuous speech recognition

    AU - Ogawa, Tetsuji

    AU - Kobayashi, Tetsunori

    PY - 2005/7

    Y1 - 2005/7

    N2 - We extend the state-observation dependencies in a Partly Hidden Markov Model (PHMM) and apply this model to continuous speech recognition. In a PHMM the observations and state transitions are dependent on a series of hidden and observable states. In the standard formulation of a PHMM, the observations and state transitions are conditioned on the same hidden state and observable state variables. Here we also condition the observations and state transitions on the same hidden states but condition the observations and state transitions on different observation states, respectively. This simple improvement to the model gives it significant flexibility allowing it to model stochastic processes more precisely. In addition, by integrating the PHMM containing this extended state-observation dependency with a standard HMM we can construct a stochastic model that we call a Smoothed Partly Hidden Markov Model (SPHMM). Results of continuous speech recognition on a newspaper read-speech have shown reductions of 10 and 24% in the error rate using the PHMM and SPHMM, respectively, compared to a standard HMM thereby displaying the effectiveness of the proposed models.

    AB - We extend the state-observation dependencies in a Partly Hidden Markov Model (PHMM) and apply this model to continuous speech recognition. In a PHMM the observations and state transitions are dependent on a series of hidden and observable states. In the standard formulation of a PHMM, the observations and state transitions are conditioned on the same hidden state and observable state variables. Here we also condition the observations and state transitions on the same hidden states but condition the observations and state transitions on different observation states, respectively. This simple improvement to the model gives it significant flexibility allowing it to model stochastic processes more precisely. In addition, by integrating the PHMM containing this extended state-observation dependency with a standard HMM we can construct a stochastic model that we call a Smoothed Partly Hidden Markov Model (SPHMM). Results of continuous speech recognition on a newspaper read-speech have shown reductions of 10 and 24% in the error rate using the PHMM and SPHMM, respectively, compared to a standard HMM thereby displaying the effectiveness of the proposed models.

    KW - Acoustic models

    KW - Continuous speech recognition

    KW - HMM

    KW - PHMM

    KW - SPHMM

    UR - http://www.scopus.com/inward/record.url?scp=20344386231&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=20344386231&partnerID=8YFLogxK

    U2 - 10.1002/scj.20315

    DO - 10.1002/scj.20315

    M3 - Article

    AN - SCOPUS:20344386231

    VL - 36

    SP - 31

    EP - 39

    JO - Systems and Computers in Japan

    JF - Systems and Computers in Japan

    SN - 0882-1666

    IS - 8

    ER -