Search

The capacity of a certain asymmetrical binary channel is studied under the following conditions. 1) Blocks of equal numbers of binary digits are used as the transmitting symbols. 2) The channel resumes its quiescent state at the beginning of each block. 3) The memory of the channel is characterized by the dependence of the noise probabilities for each digit upon the preceding digit or digits in the same block. It is shown that, by means of simple rules and with the aid of a single set of curves or a table, the calculation of the capacity can be reduced to a routine process.

Spectrographic analysis, autocorrelation, and infinite clipping are considered as methods of transforming and analyzing speech sounds with the object of obtaining simple representations without excessive loss of intelligibility. Although not mathematically equivalent in the exact sense, these methods when properly approximated are shown to provide parameters of statistical nature that are simply related. It is conjectured that the parameters describe some essential elements of speech sounds and are statistically invariant. Some experimental results are included.

A system for the analysis and synthesis speech is described. In the analyzer, speech sounds are first classified into nonturbulent and turbulent groups. The first three formants of the former group and the first three moments of the latter group constitute six significant parameters of speech spectra. By measuring the zero‐crossing densities and/or envelopes of automatically selected frequency bands, these parameters or their equivalents are extracted. To test the feasibility of using these parameters in a speech compression system, a synthesis procedure is carried out. The synthesized speech and its spectrograms are demonstrated. [The research in this paper has been made possible through support and sponsorship extended by the Electronics Research Directorate of the Air Force Cambridge Research Center, under Contract No. AF 19(604)‐1039, Item I.]

This paper describes methods for determining some statistical properties of speech sounds with the purpose of providing simple visible representations. Three such parameters are the density of zero crossings (ρ0), the density of points of zero slope (ρm), and the power ((f(t)2)Av)—all averaged over a short time interval. It can be shown that, subject to certain assumptions, these three quantities are related in rather simple ways to the moments of first few orders of the spectrogram, and the initial values of the derivatives of the autocorrelation graph. Experimental techniques involved in the determination, portrayal, and interpretation of the parameters ρ0, ρm, and (f(t)2)Av are described.

The points of zero crossing and the points of zero slope of the oscillograms of speech sounds are considered to contain the essential information for intelligibility. The intervals between zero crossings θ0, and the intervals between zero slopes θm, are plotted as points in rectangular coordinates. The ordinate of the dot is a function of θ(θ0 or θm), and the abscissa is a function of the time of occurrence t of the particular θ. The choice of these functions depends upon the types of portrayal needed for a specific analysis of speech sounds. The resulting intervalgram gives a halftone picture (consisting of dots) of speech sounds. The patterns may be proportioned to show either a detailed or general representation of the variation of the interval distribution. One type of pattern portrayed at the speech rate on a cathode‐ray oscilloscope with a screen of long persistence has been found quite similar in certain respects to the patterns obtainable by the sound spectrograph as described in the book Visible ...