Jove
Visualize
Contact Us
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

Related Experiment Videos

Frequent word section extraction in a presentation speech by an effective dynamic programming algorithm.

Yoshiaki Itoh1, Kazuyo Tanaka

  • 1Iwate Prefectural University, Sugo, Takizawa-mura, Iwate 020-0193, Japan. y-itoh@iwate-pu.ac.jp

The Journal of the Acoustical Society of America
|September 21, 2004
PubMed
Summary

This study introduces a new method to find frequently repeated words in speech data, even when general speech recognition fails. The Shift Continuous Dynamic Programming (Shift CDP) algorithm efficiently identifies these key speech sections for better search and summarization.

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

The First Autopsy-Proven Case of CSF1R-Related Leukoencephalopathy Harboring the p.Cys774Arg Mutation.

Neuropathology : official journal of the Japanese Society of Neuropathology·2026
Same author

HFA-PEFF score as a predictor of worsening heart failure after first-time catheter ablation for atrial fibrillation in patients with preclinical heart failure and preserved ejection fraction.

Frontiers in cardiovascular medicine·2026
Same author

Eosinophilic Myocarditis With Eosinophil Extracellular Traps and Charcot-Leyden Crystals.

The Canadian journal of cardiology·2026
Same author

Anti-Programmed Death-Ligand 1 Antibody-Associated Eosinophilic Granulomatosis with Polyangiitis in Non-Small Cell Lung Cancer: A Case Report.

Internal medicine (Tokyo, Japan)·2025
Same author

Direct Fast Scarlet and Congo Red Staining in Identification of Eosinophils and Their Cell-Free Granules in Cardiac Tissue.

International heart journal·2025
Same author

IgA λ-Type Schnitzler Syndrome With Edema and Numbness.

The Journal of dermatology·2025

Area of Science:

  • Speech Processing
  • Natural Language Processing
  • Computational Linguistics

Background:

  • Word frequency analysis is crucial for text searching and summarization.
  • Identifying frequent words in speech data is challenging due to specialized terminology and recognition errors.
  • Existing methods often require language models or domain-specific terms, limiting their applicability.

Purpose of the Study:

  • To develop an effective approach for automatic extraction of frequent word sections from monologue speech data.
  • To enable searching and summarization of speech datasets without prior language models or specific terms.
  • To identify key speech segments that can serve as labels or digests of the presentation content.

Main Methods:

  • The proposed method detects similar audio sections representing the same word or phrase.

Related Experiment Videos

  • Utilizes the efficient Shift Continuous Dynamic Programming (Shift CDP) algorithm for fast matching of arbitrary speech sections.
  • Employs frame-synchronous extraction to identify repeated segments in real-time.

    • Main Results:

    • Shift CDP successfully detects similar sections within speech data.
    • The algorithm accurately identifies frequent word sections in individual speeches.
    • Demonstrated effectiveness in extracting repeated sections from academic conference presentations in Japan.

    Conclusions:

    • The developed approach is domain-independent and applicable to any monologue speech.
    • Shift CDP provides a robust solution for extracting meaningful speech segments without prior knowledge.
    • The extracted frequent word sections can serve as valuable speech labels or summaries.