SPEECH BANDWIDTH REDUCTION.

Abstract

This report presents experimental results, comclusions, and recommendations resulting from research - conducted on a speech bandwidth reduction technique based on an electrical analog of the human hearing mechanism. This research centered around the Single Equivalent Formant (SEF) technique whereby it is hypothesized that intelligent speech could be transmitted with three parameters: SEF frequency, amplitude, and pitch. This was partially proven in that the intelligence (RT word score) was measured at 75 percent for experienced speakers and listeners. However, the machine-like quality of the synthesized speech proved to negate the system's use with all but highly trained operators. Three other systems, including resynthesizing three formant speech from the SEF parameter using storage, were investigated. The system that was finally decided upon as the closest to meeting the goals is a two formant, three amplitude, and pitch system with 120 Hz bandwidth. This system has a measured RT score of 85 percent and the quality is adjudged to be quite human-like. Transmission of the parameters is accomplished.

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1966
Accession Number
AD0801360

Entities

People

  • David M. Jurenko
  • John P. Schultz
  • Richard Abt

Tags

DTIC Thesaurus Topics

  • Amplitude
  • Bandwidth
  • Frequency

Readers

  • Speech Processing/Speech Recognition.
  • Systems Analysis and Design