An experiment for the representation of voice signals by means of an algorithm based on a product of homotopic correlations
Abstract
An algorithm derived from a model of the human auditive system is presented and used for the analysis of voice signals. The algorithm is defined as a transformation, involving the homotopic products along the model delay line and weighting functions related to the receptor memories. The model consisted of 65 receptors equally spaced along the two sides of a delay line with a delay constant of 0.1 msec between receptors. The time weighting functions are exponential and were chosen to obtain an attenuation of 20 db in 25 msec. It is shown that the homotopic correlation algorithm can follow variations in signal strength and provides adequate information on the characteristics of small signal segments.
 Publication:

NASA STI/Recon Technical Report N
 Pub Date:
 September 1979
 Bibcode:
 1979STIN...8127378B
 Keywords:

 Algorithms;
 Data Correlation;
 Homotopy Theory;
 Speech Recognition;
 Voice Data Processing;
 Auditory Perception;
 Delay Lines;
 Mathematical Models;
 Phonemics;
 Verbal Communication;
 Weighting Functions;
 Communications and Radar