Browse Prior Art Database

Detection of Excitation Times of the Mouth Cavities, and Measurement of Fundamental Voice Frequency

IP.com Disclosure Number: IPCOM000078918D
Original Publication Date: 1973-Apr-01
Included in the Prior Art Database: 2005-Feb-26
Document File: 5 page(s) / 79K

Publishing Venue

IBM

Related People

Nemeth, A: AUTHOR

Abstract

Voice signals are coded in pulse-code modulation (PCM). See asterisks, Fig. 1. Absolute values d1, d2, etc. of the different amplitudes of the voice signal are first obtained between two consecutive samples. Values d1, d2, ... etc. are increasing, and become high about the beginning of the excitation period. About the maximum excitation time, they are decreasing, and become low. Secondly, each difference d(i) is added to the sum of the absolute values of the preceding differences. Thus, values S(i) are obtained: S1 = d1; S2 = d1 + d2; Sx = d1 + d2 + d3 + ... dx.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 27% of the total text.

Page 1 of 5

Detection of Excitation Times of the Mouth Cavities, and Measurement of Fundamental Voice Frequency

Voice signals are coded in pulse-code modulation (PCM). See asterisks, Fig.
1. Absolute values d1, d2, etc. of the different amplitudes of the voice signal are first obtained between two consecutive samples. Values d1, d2, ... etc. are increasing, and become high about the beginning of the excitation period. About the maximum excitation time, they are decreasing, and become low. Secondly, each difference d(i) is added to the sum of the absolute values of the preceding differences. Thus, values S(i) are obtained: S1 = d1; S2 = d1 + d2; Sx = d1 + d2 + d3 + ... dx.

Different values with respect to time are in Fig. 2. portion A of the curve, corresponds to the times following the maximum excitation time of the mouth cavities; portion B corresponds to the ends of the excitation periods during which values d(i) are low and sometimes, zero; points C correspond to the excitation times of the vocal cords at the beginning of each voice period.

The sudden change in the slope of the curve accompanied with a change in the curvature is retained to characterize the excitation times of the mouth cavities. At each point on the curve of the sums of the differences in Fig. 2, a measurement is made of the difference between two slopes, the lower slope PI and higher slope PS. This supplies information relative to the sudden change in the slope, and also the curvature at each of its points.

To avoid errors introduced by irregular values, detection of change in the curvature and the sudden variations in the slope is done by evaluating, for each point on the curve, the value of slope PI and that of slope PS on a same number N of values S(i) preceding or following value S(i), which corresponds to that point on the curve for which the curvature is calculated. Number N is chosen to permit detecting times corresponding to the opening of the vocal cords of any person. Vocal cords having the closest opening times, such as the highest pitched voices of women, are opened every 3 milliseconds. The values of number N are chosen so as to be lower than about 3 milliseconds, making it possible to have first portion A having an abrupt slope and second portion B having a smooth slope available at any point on the curve for the calculation of the curvature. Detection of the times when the sudden variations in the slope occur is carried out for each point "i" of the curve, by making the difference between value S(i) at that point and value S(i-20) at point "i-20" for the obtainment of PI(i), on the one hand, and the difference between values S(i+20) and S(i) for the obtainment of pS(i), on the other hand, and by subtracting PS(i) from PI(i). For each point "i", the following calculation is carried out: PI(i) - PS(i) = [S(i) - S(i-20)] - [S(i+20) - S(i)] = D(i).

Fig. 3 shows the curve for the mean values obtained from points corresponding to the value of D(i) = PI(i) - PS(i...