F-16 Speaker-Independent Speech Recognition System Using Cockpit Commands (70 Words)

Abstract

A system is developed to achieve speaker independent isolated speech recognition. The system uses LPC Spectrum, Formants, and Frication Frequency as a feature set. Dynamic programming is applied for distance calculations. The fundamental design concept is to create a universal template for multiple speakers. A new algorithm, which combines the vocabularies of several speakers to produce one optimal template, is incorporated into the system. An advanced speech analysis tool called SPIRE provides the computational functions required to extract appropriate features. Seventy words, from the list of F-16 cockpit commands, are selected as a vocabulary of the system. The use of a merged template based on three of the feature sets achieves an accuracy of 99 percent. The system is implemented in list programming on Symbolics 3600 series computer. Keywords: Dynamic programming, Multiple feature sets, Merged template, LISP (list Processing Language), Data processing, Theses.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1988
Accession Number
ADA203177

Entities

People

  • Peter Y. Kim

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Biomedical
  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Automated Speech Recognition
  • Automatic Gain Control
  • Band Spectra
  • Computer Programming
  • Data Processing
  • Electrical Engineering
  • Engineering
  • Feature Extraction
  • Fuzzy Sets
  • Human Factors Engineering
  • Information Processing
  • Language
  • Lisp Programming Language
  • Pattern Recognition
  • Recognition
  • Two Dimensional
  • Word Recognition

Fields of Study

  • Computer science

Readers

  • Computational Linguistics
  • Computational Modeling and Simulation
  • Speech Processing/Speech Recognition.

Technology Areas

  • AI & ML
  • AI & ML - Information Retrieval