The Intelligibility of Words, Sentences, and Continuous Discourse Using the Articulation Index

Abstract

The purpose of this research was to investigate the effect of message redundancy upon intelligibility. The original methodology for the Articulation Index (AI) French and Steinberg, J. Acoust. Soc. Am., 19, 90-119, 1947 was used to examine the relation between words, meaningful sentences and continuous discourse (CD). One primary consideration was to derive the relations between the three speech types with tightly controlled, highly repeatable experimental conditions such that any difference between them could be attributed solely to inherent contextual differences. One male speaker recorded 616 monosyllabic words, 176 meaningful speech perception in noise (SPIN) sentences and 44 seventh-grade reading level CD passages. Twenty-four normal hearing subjects made intelligibility estimates of the CD and sentences and identified words at each of 44 conditions of filtering and signal-to-noise ratio. The sentence intelligibility scores and continuous discourse intelligibility scores plotted versus the AI (transfer function) were within 0.05 AI of each other. The word recognition scores were considerably lower for equivalent AI values of both sentences and CD.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1992
Accession Number
ADA256781

Entities

People

  • R. A. Depaolis

Organizations

  • Pennsylvania State University

Tags

Communities of Interest

  • Advanced Electronics
  • Human Systems

DTIC Thesaurus Topics

  • Automated Speech Recognition
  • Communication Systems
  • Computer Programs
  • Digital Signal Processing
  • Filtration
  • Frequency Bands
  • Hearing Loss
  • Identification
  • Intelligibility
  • Language
  • Perception
  • Recognition
  • Speech
  • Standards
  • Test Methods
  • Transfer Functions
  • Word Recognition

Readers

  • Computational Linguistics
  • Regression Analysis.
  • Speech Processing/Speech Recognition.