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

Experiments in dysarthric speech recognition using artificial neural networks

G Jayaram1, K Abdelhamied

  • 1Department of Biomedical Engineering, Louisiana Tech University, Ruston 71272, USA.

Journal of Rehabilitation Research and Development
|May 1, 1995
PubMed
Summary

Artificial neural networks (ANNs) show promise in recognizing dysarthric speech. These advanced systems outperformed human listeners and commercial software in speech recognition tasks.

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

Study of outcomes of cemented hemiarthroplasty for comminuted intertrochanteric fractures in the elderly.

Journal of orthopaedics·2023
Same author

Can Mobile Health Improve Depression Treatment Access and Adherence Among Rural Indian Women? A Qualitative Study.

Culture, medicine and psychiatry·2020
Same author

Polarity and timing-dependent effects of transcranial direct current stimulation in explicit motor learning.

Neuropsychologia·2011
Same author

Cytology of mucinous carcinoma of breast: a report of 28 cases with histological correlation.

The Malaysian journal of pathology·2005
Same author

Evaluation of fine needle aspiration cytology of salivary glands: an analysis of 141 cases.

The Malaysian journal of pathology·2002
Same author

Electroconvulsive therapy for depression in a cardiac transplant patient.

Psychosomatics·2001

Area of Science:

  • Speech processing
  • Artificial intelligence
  • Biomedical engineering

Background:

  • Dysarthric speech presents significant challenges for automated recognition due to high variability.
  • Developing robust speech recognition systems for individuals with speech impairments is crucial for communication accessibility.

Observation:

  • Two artificial neural networks (ANNs) were developed and trained using dysarthric speech data.
  • Input features included Fast Fourier Transform (FFT) coefficients and formant frequencies, with variations explored.
  • Performance was benchmarked against human intelligibility ratings and a commercial speech recognition system.

Findings:

  • The ANNs demonstrated a strong ability to recognize dysarthric speech, even with its inherent variability.
  • Networks utilizing FFT coefficients and formant frequencies achieved high recognition rates.

Related Experiment Videos

  • Both ANN models significantly outperformed human listeners and the Introvoice commercial system in accuracy.

    • Implications:

    • This research highlights the potential of ANNs for improving speech recognition technology for individuals with dysarthria.
    • The findings suggest that ANNs can offer a more reliable solution compared to existing methods.
    • Further development could lead to enhanced communication tools for those with speech disorders.