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

Multi-expert and hybrid connectionist approach for pattern recognition: speaker identification task

Y Bennani1

  • 1C.N.R.S., L.I.P.N. URA-1507, University of Paris-Nord, Villetaneuse, France.

International Journal of Neural Systems
|September 1, 1994
PubMed
Summary

This study introduces a novel modular/hybrid connectionist system for speaker identification. The system achieved perfect identification accuracy, outperforming existing methods in both speed and precision.

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

Achievement of behavioral efficacy and improved potency in new heterocyclic analogs of benzofuran H3 antagonists.

Inflammation research : official journal of the European Histamine Research Society ... [et al.]·2005
Same author

The medicinal chemistry of novel H(3) antagonists.

Inflammation research : official journal of the European Histamine Research Society ... [et al.]·2004
Same author

Molecular and pharmacological characterization of the monkey histamine H(3) receptor.

Inflammation research : official journal of the European Histamine Research Society ... [et al.]·2004
Same author

Features selection and architecture optimization in connectionist systems.

International journal of neural systems·2001
Same author

A modular and hybrid connectionist system for speaker identification.

Neural computation·1995
Same author

[Segmental small intestine transplantation in dogs: comparison between a jejunal graft and an ileal graft].

Annales de chirurgie·1992

Area of Science:

  • Artificial Intelligence
  • Speech Processing
  • Machine Learning

Background:

  • Speaker identification is complex, often limited by scarce training data.
  • Modular connectionist systems offer better generalization and incorporate prior knowledge.
  • Hybrid systems can combine connectionist and non-connectionist approaches, like Hidden Markov Models (HMMs).

Purpose of the Study:

  • To present and evaluate a modular/hybrid connectionist system for text-independent speaker identification.
  • To demonstrate the effectiveness of modularity and hybrid approaches in speaker recognition.
  • To address challenges of limited training data in speaker identification tasks.

Main Methods:

  • Developed a modular/hybrid connectionist architecture.

Related Experiment Videos

  • Integrated several connectionist modules with a Hidden Markov Model (HMM) module.
  • Tested the system on 102 speakers from the DARPA-TIMIT database.

    • Main Results:

    • Achieved perfect speaker identification (100% accuracy) on the test population.
    • Demonstrated superior performance compared to multivariate auto-regressive models in accuracy and speed.
    • Results are among the best for text-independent speaker identification systems of this scale.

    Conclusions:

    • The modular/hybrid connectionist approach is highly effective for text-independent speaker identification.
    • This system offers significant advantages in accuracy, speed, and adaptability.
    • The design facilitates easy integration of new speakers into the system.