Jove
Visualize
Contact Us

Related Experiment Videos

Linear prediction analysis of speech based on a pole-zero representation

B S Atal, M R Schroeder

    The Journal of the Acoustical Society of America
    |November 1, 1978
    PubMed
    Summary

    This study introduces a novel pole-zero model for speech analysis, improving spectral accuracy over traditional all-pole linear prediction models. The noniterative method accurately characterizes speech, especially nasal sounds.

    Related Concept Videos

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Guidelines for care of the dying patient: an interdisciplinary effort.

    Continuum (Society for Social Work Administrators in Health Care)·1997
    Same author

    Speech technology in 2001: new research directions.

    Proceedings of the National Academy of Sciences of the United States of America·1995
    Same author

    Crystallization and preliminary X-ray diffraction studies of recombinant barley lectin and pro-barley lectin.

    Journal of molecular biology·1993
    Same author

    Determination of the functional elements within the vacuolar targeting signal of barley lectin.

    The Plant cell·1993
    Same author

    Colocalization of barley lectin and sporamin in vacuoles of transgenic tobacco plants.

    Plant physiology·1993
    Same author

    Isolation and characterization of pro-barley lectin expressed in Escherichia coli.

    Protein expression and purification·1992
    JoVE
    x logofacebook logolinkedin logoyoutube logo
    ABOUT JoVE
    OverviewLeadershipBlogJoVE Help Center
    AUTHORS
    Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
    LIBRARIANS
    TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
    RESEARCH
    JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
    EDUCATION
    JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
    Terms & Conditions of Use
    Privacy Policy
    Policies

    Area of Science:

    • Speech processing
    • Digital signal processing
    • Acoustics

    Background:

    • Traditional linear predictive coding (LPC) uses an all-pole model, which struggles to accurately represent speech spectra, particularly for nasal and nasalized sounds.
    • Accurate spectral envelope modeling is crucial for speech analysis and synthesis.

    Purpose of the Study:

    • To present a new method for speech characterization using a pole-zero model.
    • To overcome the limitations of all-pole models in representing complex speech spectra.

    Main Methods:

    • Constructing an impulse response from the speech signal, incorporating glottal, vocal tract, radiation, and recording system effects.
    • Utilizing multiple stages of all-pole linear predictive coding (LPC) analysis to derive the impulse response.
    • Determining pole-zero parameters by solving simultaneous linear equations, offering a noniterative approach.

    Main Results:

    • The proposed pole-zero modeling method provides a more accurate description of speech spectra compared to all-pole models.
    • The noniterative nature of the method makes it suitable for automatic speech analysis.
    • Validation on real speech data shows close agreement between pole-zero model spectra and direct Fourier analysis spectra.

    Conclusions:

    • The pole-zero model offers a significant improvement for speech spectral analysis, especially for challenging sounds like nasals.
    • The noniterative, equation-solving approach facilitates efficient and automated speech analysis.
    • This method enhances the accuracy of speech representation in digital signal processing applications.

    Related Experiment Videos