OPTIMUM SPEECH SIGNAL MAPPING TECHNIQUES

Abstract

Investigations of the analysis of speech in terms of a fixed exponential function series were carried out. The analysis-synthesis processing was performed via digital computers operating with digitized speech. The results indicate that the representation used is an efficient one for speech waveform analysis and that the information content of the speech is preserved when phase information is eliminated. The spectral coefficients after phase elimination are not found to be an efficient representation for the amplitude spectrum. Other results show that the method of analysis is not limited to the speech of one individual. Analytical studies indicate that it is possible to optimize the method of analysis to essentially perfect its efficiency for speech waveform analysis. Normalized, phase eliminated, spectral patterns derived for ten vowel utterances by five talkers indicate the feasibility of performing both automatic vowel and/or automatic speaker recognition using the orthonormal coefficient data.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 28, 1962
Accession Number
AD0296333

Entities

Organizations

  • Sylvania Electric Products

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Air Force
  • Analysis Of Variance
  • Automated Speech Recognition
  • Contracts
  • Correlation Analysis
  • Covariance
  • Cross Correlation
  • Data Science
  • Electronic Warfare
  • Identification
  • Information Science
  • Magnetic Tape
  • Specifications
  • Speech Analysis
  • Square Roots
  • Statistical Analysis
  • United States

Fields of Study

  • Engineering

Readers

  • Image Processing and Computer Vision.
  • Speech Processing/Speech Recognition.
  • Statistical inference.

Technology Areas

  • AI & ML