Volume 41, Issue 6, January 1967

Transversal Motion in Transformer Laminations as a Cause of Noise
View Description Hide DescriptionTransversal (out‐of‐plane) motion of transformer laminations has been noted by investigators of transformer noise, but it can be explained readily neither by electromagnetic forces between poles nor by magnetostriction. A new approach to the theory of transversal vibration is presented that suggests a cause of this type of lamination vibration and noise, namely, any magnetic path that causes magnetic flux to cross an air gap provides a force mechanism resulting in significant out‐of‐plane motion. This approach is particularly applicable to ballast transformers, in which air gaps, lamination slots, or leakage flux are commonly used. The theory furnishes expressions for the expected values of lamination displacement and velocity. A comparison with experimental data for one set of parameters is given.

Matrix Formulation in Acoustical Analysis of Mechanically Driven Fluid Systems
View Description Hide DescriptionMechanically driven fluid systems are inherent acoustic‐wave generators because of their motion and should thus be classified as time‐variant. As a direct consequence, the acoustic wave in the system will consist of a series of discrete components at the fundamental time‐varying frequency and harmonics thereof, and will lend itself to Fourier‐series representation. Previous works assumed the moving part to be a constant‐current source in a time‐invariant acoustic equivalent circuit as a convenient approximation. This work, however, describes the matrix formulation of a solvable linearized time‐variant equivalent circuit for these systems. By use of matrix theory, a general solution with better accuracy for acoustic wave throughout a fluid system is derived. The solution is within the capability of a large digital computer. Later, the simplification of the time‐variable element into a constant‐current source approximation is described.

Influence of Compressor Action on Speech Intelligibility
View Description Hide DescriptionThe effects of compressor action, which reduces the long‐term dynamic range of speech, on intelligibility of monosyllabic words presented in quiet was investigated. Three ratios of change in input level to change in output level were used, namely, 1‐ to 1‐dB, 2‐ to 1‐dB, and 3‐ to 1‐dB. Peak powers of the speech signals were adjusted for these three conditions of reproduction so as to achieve presentations of 0, 8, 16, and 24 dB sensation level. The interphonemic dynamic range of the speech signal was the variable under study. Performance of the normal hearers, patients with labyrinthine hydrops, labyrinthine otosclerosis, and presbycusis improved only slightly during either condition of compression. Moreover, the subjects with hearing loss did not achieve any greater benefit from compressor action than did normals. These results warrant the conclusion that neither the 2‐ to 1‐dB nor the 3‐ to 1‐dB compression ratio offered any important advantage over 1‐ to 1‐dB reproduction when comparison was made only in terms of intelligibility at a given sensation level of the output signal.

Machine‐Aided Formant Determination for Speech Synthesis
View Description Hide DescriptionA semi‐automatic analysis‐synthesis scheme that can be viewed as a “manual formant vocoder” is described. A human operator makes decisions about formant positions on processed speech data. The parameters which result from the operator decisions are used to control a four‐pole parallel synthesizer.Speech processed by the system had an error rate of 4.2% for vowels and 16.9% for consonants.

Parameter of Nonlinearity in Fluids. III. Values of Sound Velocity in Liquid Metals
View Description Hide DescriptionThe speed of sound is reported for liquid metals(mercury,indium,bismuth,sodium, and tin) as a function of pressure (0 to 150 atm) and temperature (up to 100°K above the melting point). The result and data are used to obtain expressions for the derivatives of the sound speed with respect to temperature and pressure. These coefficients are then used to determine the parameter of nonlinearity in terms of the ratio B/A of the first two coefficients of the adiabatic expansion of pressure in terms of the density.

Mode Conversion in the Long‐Wavelength Limit
View Description Hide DescriptionConversion of energy from the first‐order longitudinal to the first‐order transverse mode is observed in the long‐wavelength limit for guided‐wave propagation in solid cylinders. This conversion process increases the attenuation of a multiply reflected longitudinal pulse and occurs in a localized region near any deviation from cylindrical symmetry in the waveguide.

Cross Section of a Circular Cylinder with an Impedance‐Loaded Strip along a Generatrix
View Description Hide DescriptionAn expression is obtained for the monostatic cross section of a circular cylinder when the cylinder sustains a surface impedance load which is constant over its surface, except possibly on a narrow strip along a generatrix. To get this expression, an integral formulation of the problem is obtained, and a discrete matrix approximation is taken. The matrices are diagonalized, and the problem is reduced to a scalar one. Then, because the finite series in the resulting expressions are discrete forms of well‐known integrals, these integrals are replaced by their exact values. The resulting expression is well suited to study the effect of strip width and strip impedance load on the cross section of a circular cylinder. The expression is then used to study the effect which certain nonhard strips have on the cross section of the hard cylinder.

Pressure Correlation for a Cylinder in a Diffuse Sound Field
View Description Hide DescriptionA theoretical calculation is given of the surface‐pressure correlation function for a rigid, infinitely long cylinder in a three‐dimensional diffuse sound field (i.e., a sound field in which waves having all frequencies impinge on the cylinder from all directions). The analysis includes the effects of scattering of the soundwaves by the cylinder. A general expression is derived that gives surface pressure spatialcorrelation as a function of both circumferential and longitudinal separation for any arbitrary distribution of sound amplitude versus frequency. With the aid of a computer, numerical results are obtained for the case of a narrow‐band acoustic noise field. These results show a decrease in spatialcorrelation with increasing circumferential and longitudinal separation , as well as with increasing frequency. The results also indicate that scattering by the cylinder affects the spatialcorrelation only slightly in the longitudinal direction, but that it affects it considerably in the circumferential direction.

Frequency Dependence of Reverberation in the Ocean
View Description Hide DescriptionData on acoustic reverberation, collected at 12 stations in the Pacific Ocean, are presented to show gross characteristics of the volume reverberation as a function of depth at two frequencies, 10 and 25 kHz. The experimental equipment consisted of a cable‐connected, horizontally directional echo‐ranging instrument package, operating in conjunction with a shipboard digital data recorder. Of interest are the comparisons of relative scattering coefficients measured simultaneously at two frequencies. These comparisons indicate significant changes of scatterer composition with depth.

Epstein Normal‐Mode Model of a Surface Duct
View Description Hide DescriptionEquations are given for the normal‐mode calculation of the sound field in a surface duct. The velocity‐depth profile is represented as , a form first used by Epstein for calculating the reflection of radio waves from a transition layer. A comparison of calculated and experimental measurements of propagation loss shows good agreement between theory and experiment.

Shallow‐Water Propagation under Downward‐Refraction Conditions
View Description Hide DescriptionThe important features of shallow‐water sound transmission under downward‐refraction conditions are interpreted in terms of ray theory and are discussed relative to some specific 30‐f (fathom) propagation‐loss results. The propagation results are shown to depend directly upon the velocity profile and the bottom loss. A bottom‐loss value of 1.3 dB per bounce was inferred from the measurements and is shown to compare favorably with theoretical values computed from bottom‐impedance information. Observed pulse shapes are then compared with theoretical pulse shapes synthesized from the combined ray and bottom‐loss models. In addition, the computed propagation‐loss values for the various paths are presented as a function of source angle in order to illustrate the vertical directivity introduced by the environment. Finally, the effect of this vertical directivity on shallow‐water propagation measurements is demonstrated by comparing propagation‐loss results for directional and explosive sources.

Specular Scattering of Underwater Sound from a Wind‐Driven Surface
View Description Hide DescriptionA wind‐driven water surface possessing the gross characteristics of a simple sea (near‐Gaussian distribution of slopes and a 1.53 ratio of the upwind‐downwind slope to the cross slope) has been studied acoustically by analyzing the specular scattering of a broad range of frequencies at normal incidence surface roughness √g = 4πσ cosθ/λ ranging from 0.56 to 5.1). Theory, based on the Helmholtz integral and the Kirchhoff approximation, modified for beam diffraction, is used to provide values of the root‐mean‐square surface height from specular scatter at small roughness and of the root mean square slope from the asymptotic results at large roughness. The predicted exponential decrease of intensity with increasing but low value of roughness is corroborated even for wavelengths as large as the surface correlation distance, that is, far beyond the restrictions of the Kirchhoff assumption. The acoustically determined value of the root‐mean‐square slope is 17% less than the optically measured value. The form of the distribution of specularly scattered pressure is found to vary from near Gaussian distribution for the relatively smooth surface to near‐Rayleigh distribution for the rough surface. At any instant, the relative specularly scattered intensity decreases monotonically with increasing frequency to a minimum value at √g≃2.3 from which it is possible to get a second estimate of the root mean square height: above √g∼2.3 the instantaneous scattered intensity is a nonstationary oscillating function of frequency, varying about the asymptotic value.

Angular Distribution of Lower Room Modes
View Description Hide DescriptionThe angular distribution of the lower frequency modes in a rectangular room is calculated for various room dimensions. Applying the results to reverberation‐room measurements, it is shown that a sample placed with its surface normal to a long dimension of the room experiences the most diffuse field. Various methods are proposed to modify the existing techniques of absorption measurement, where it usually happens that the sample is placed with its surface normal to the shortest axis, thus experiencing the least diffuse field.

Mechanistic Aspects of Hearing
View Description Hide DescriptionThe results of a number of apparently diverse experiments on human hearing can be summarized via a simple model. Two mechanisms are required: One element emits unit responses whose number is proportional to the instantaneous amplitude of a sound. The other is a selector mechanism, analogous to a resonant circuit, that integrates the responses. Correspondingly, just two parameters are involved: the threshold of hearing and the “Q” of the selector mechanism. The system of equations representing the model makes it possible to derive the necessary parameters from the results of a number of different types of experiments. The theory of the selector mechanism is based upon a recent study of the limits of performance of a system capable of storing oscillatory energy reversibly, but the remainder of the theoretical description makes use of ordinary communication theory. The discrimination ability of the normal ear is shown to be proportional to the one‐fourth power of the level above threshold, and to relate closely to the sensation of loudness. Many features of the behavior of normal and impaired ears can be predicted by application of familiar network theory to the parameters derived from the model. The properties of the model may be used as criteria in the choice of experiments to develop explicit data on the way the ear performs its function.

Additivity of Masking
View Description Hide DescriptionSuppose, in separate experiments, the level of noise and the level of a sinusoid are adjusted so that each produces the same amount of masking of a gated sinusoidal signal. If now the noise and sine‐wave masker are combined, how much must the level of the signal be increased to achieve the same level of masking effectiveness as before? Few conditions require only a 3 dB increment of the signal. It is argued that an increment greater than 3 dB implies that the processing of the signal when masked by noise and when masked by a sine wave are different.

Pitch of Complex Tones
View Description Hide DescriptionThe question of whether the pitch of complex tones is based either on the frequency of the fundamental or on the periodicity of the sound as a whole was studied. Fifteen naïve subjects compared the stimuli and had to decide which of the two stimuli, presented successively during 200‐msec intervals, was higher in pitch. The results showed that for m = 1 (fundamental of B was 10% lower in frequency than that of A, and all other harmonics were 10% higher in frequency than A) the pitch changed according to the harmonics for f < 1400 Hz and according to the fundamental for f> 1400 Hz. For m = 2, 3, and 4, the corresponding transition frequency was about 700, 350, and <125 Hz, respectively. This holds for both a_{n} = 1 and a_{n} = 1/n. The experiments support the view that the pitch of complex tones is based on periodicity rather than on frequency. The description of the experiments is preceded by a historical review.

Quality of Violin Vibrato Tones
View Description Hide DescriptionTypical violintones with vibrato were recorded on magnetic tape. From a special analysis, synthetic tones were constructed that were very difficult to distinguish from real tones. The analysis showed that not only the frequency was varying up and down about sit times per second, but there were three other important characteristics varying at this same rate. All of the harmonics had the same variation of the frequency level, in cents, from the note being played. There was also an intensity‐level variation of the harmonics at the same vibrato rate, but the intensity‐level variation was greatly different for different harmonics. Also, the intensity of some of the harmonics would be rising while that of others would be falling during the vibrato period. Thus, a curve showing the relative intensity level of the harmonics at any instant changed back and forth during the vibrato cycle. The intensities of the sympathetic tone coming from the open strings also varied either at the vibrato rate or twice the vibrato rate, while the frequencies of these tones remain constant and correspond to the natural frequencies of the fundamental and harmonics of the open strings. The after‐ring is due principally to these transient tones. This effect is negligible when the frequencies of the fundamental and its harmonics are not close to those of the open string.
 LETTERS TO THE EDITOR


Attenuation Provided by Fingers, Palms, Tragi, and V51R Ear Plugs
View Description Hide DescriptionMeasurements of the attenuation of noise provided by use of the fingers, palms, tragi, and V51R ear plugs are presented for nine test tones. The results of the study show that the tragi provide the best attenuation, the palms and fingers slightly less attenuation, and the ear plugs the least attenuation.

Random Efficiency of Second‐Order Gradient Microphones
View Description Hide DescriptionThe random efficiency of the second‐order Limaçon transducer family, , is computed and minimum random efficiency of 0.125 is determined to occur when either m or n is and the other is 0.

Asymmetries in the Cumulative Probability Distributions of the Speech Waveform Instantaneous Amplitude
View Description Hide DescriptionAsymmetries in the cumulative probability distribution of the speech waveform instantaneous amplitudes are studied. It is shown that the asymmetries which occur at large amplitude values are accompanied by a difference in the mean and median values of the distributions. Microphone characteristics and filtering reduce the gross asymmetry of the larynx fundamental waveform, resulting in a reduction in the over‐all asymmetry.
