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

Optimizing the reliability of speech recognition scores

S A Gelfand1

  • 1Queens College and the Graduate School of the City University of New York, Flushing 11367, USA. sagelfand@qc.edu

Journal of Speech, Language, and Hearing Research : JSLHR
|October 15, 1998
PubMed
Summary

Clinicians need reliable speech recognition tests. This study introduces a 450-item phonemic scoring method using computer-presented words, achieving optimal reliability for accurate speech recognition assessment.

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

[Staged treatment of patient with combined trauma with diffuse peritonitis and numerous complications].

Vestnik khirurgii imeni I. I. Grekova·2012
Same author

Long-term recovery and no recovery from the auditory deprivation effect with binaural amplification: six cases.

Journal of the American Academy of Audiology·1995
Same author

Acoustic reflex threshold tenth percentiles and functional hearing impairment.

Journal of the American Academy of Audiology·1994
Same author

Apparent auditory deprivation in children: implications of monaural versus binaural amplification.

Journal of the American Academy of Audiology·1993
Same author

Functional components and resolved thresholds in patients with unilateral nonorganic hearing loss.

British journal of audiology·1993
Same author

Organic thresholds and functional components in experimentally simulated exaggerated hearing loss.

British journal of audiology·1993

Area of Science:

  • Audiology
  • Speech-Language Pathology
  • Psychometrics

Background:

  • Speech recognition testing faces a trade-off between test length and reliability.
  • Statistical models suggest a large number of scorable items (~450) are needed for optimal test reliability.

Purpose of the Study:

  • To develop a reliable speech recognition test meeting statistical requirements.
  • To maintain key features of established speech recognition tests.

Main Methods:

  • An interactive computer program presented CNC words in 50 three-word groups.
  • Phonemic scoring was employed, yielding 450 scorable items.
  • Normal performance was defined by presentation level and signal-to-noise ratio.

Main Results:

Related Experiment Videos

  • The developed approach achieved reliability consistent with binomial model predictions for both normal-hearing and sensorineural-loss groups.
  • Phoneme scores explained 99% of word score variance across typical clinical performance ranges.
  • Test-retest reliability was validated in 200 participants.

Conclusions:

  • A computer-based, phonemically scored test with 450 items provides optimal reliability for speech recognition assessment.
  • Phonemic scoring allows for expression of results as equivalent word recognition scores, enhancing clinical utility.