Prosodic Aids to Speech Recognition. 3. Relationships between Stress and Phonemic Recognition Results

Abstract

A strategy is being implemented for acoustic aspects of speech recognition, whereby prosodic features are used to detect boundaries between phrases, then stressed syllables are located within each phrase, and a partial distinctive features analysis is done within stressed syllables. Several algorithms are being investigated for locating stressed syllables in continuous speech. Preliminary experiments have shown some success in locating sibilants and determining their places of articulation. Partial distinctive features analysis on stressed vowels has been attempted. Location of stop consonants and sibilants, and sibilant place of articulation determination have been more successful in stressed syllables. Studies are being conducted on the relative success of vowel and obstruent categorizations in stressed, unstressed, and reduced syllables. These studies in segmental analysis, and companion studies in stress perception and automatic location of stressed syllables, are being conducted on 31 ARPA Sentences, but later work will be based on new speech texts now being designed.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 21, 1973
Accession Number
AD0767392

Entities

People

  • Mark F. Medress
  • Toby E. Skinner
  • Wayne A. Lea

Organizations

  • Sperry Corporation

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Automated Speech Recognition
  • Boundaries
  • Computations
  • Computer Programs
  • Computers
  • Detectors
  • False Alarms
  • Fast Fourier Transforms
  • High Energy
  • Integrals
  • Language
  • Linguistics
  • Operating Systems
  • Recognition
  • Research Facilities
  • Speech Analysis
  • Warning Systems

Readers

  • Speech Processing/Speech Recognition.

Technology Areas

  • AI & ML