MEETINGS AND CONFERENCES

SCIENTIFIC CONFERENCE ON ELECTROACOUSTICS

From July 1 to 5, 1955, a scientific conference on electroacoustics was held in Kiev, convened by the Acoustics Commission of the Academy of Sciences of the USSR, the Acoustics Institute of the Academy of Sciences of the USSR, and the Kiev Order of Lenin Polytechnic Institute.

At the conference 17 papers were heard, most of which dealt with current problems of modern electroacoustics—questions of sound propagation and radiation, audibility and the measurement of nonlinear distortions in sound reproduction, and the theory and experimental study of sound carriers and magnetic recording. The conference listened with particular attention to papers on new methods and instruments for a number of electroacoustic measurements, as well as to a paper on the biological action of ultrasound.

A discussion was opened on a number of the papers. The conference adopted a resolution on questions concerning the development of individual directions in the field of electroacoustics, as well as recommendations concerning the production of electroacoustic measuring equipment.

L. M. Brekhovskikh, in his paper “The Present State of the Theory of the Propagation of Sound Waves,” gave a broad survey of work carried out in this field in the USSR and abroad over the past 10–15 years. Dwelling on questions of calculating the field of a concentrated radiator near the boundary between two media, he noted the erroneous ideas of Zenek, according to which the sound pressure, with distance from the radiator along an absorbing boundary, decreases exponentially with distance. The speaker and, independently, Ott showed rigorously that in reality the sound pressure decreases inversely proportionally to the square of the distance. This conclusion was also obtained in the paper by a very graphic approximate method. The developed theory makes it possible to calculate correctly the propagation of sound waves over the heads of spectators in an auditorium and to solve other practically important problems.

In the case where the wave incident on the boundary is not a plane wave but a diverging one, and the speed of sound in the lower medium is greater than the speed of sound in the upper one, a lateral wave appears in the latter (in seismology, a “head wave,” or a “creeping wave”), which is a peculiar manifestation of total internal reflection in the lower medium. The occurrence of the lateral wave is easily traced in the propagation of a sound pulse. A pulse propagating through the lower medium overtakes the pulse propagating in the upper medium and produces radiation into the upper medium at the boundary in the form of a lateral wave. Although for this wave the transition to the approximation of geometrical optics loses its meaning, nevertheless

many of its properties can be understood from ray representations. In particular, the position of the wave front is determined on the basis of Fermat’s principle as the geometric locus of points for which the total length of the “optical” paths of rays propagating partly in the upper and partly in the lower medium is minimal. It is interesting that, if the displacement of the rays upon reflection from the boundary is taken into account, then an expression can be obtained for the amplitude of the lateral wave as a function of distance.

Next L. M. Brekhovskikh gave a survey of data obtained in recent years on the propagation of waves in layers, noting that at present there is a complete theory of the propagation of sound waves in a system of plane-parallel layers with arbitrary boundaries of separation. In such a medium a single system of normal waves is established, whose basic characteristics—the phase velocity of propagation along the layer and the attenuation coefficient—can be found from the so-called dispersion equation. The excitation coefficients of each of the normal waves are determined by the position of the radiator.

The speaker then dealt with questions of the theory of super-long-range propagation of sound in the underwater sound channel—a phenomenon discovered in the USSR by L. D. Rozenberg and explained by the speaker in 1946, and in America, according to a postwar publication, as early as the war years. A sound waveguide also exists in the atmosphere, where it is formed owing to a temperature inversion at altitudes of the order of 30–50 km. Since the width of sound waveguides is, as a rule, very large in comparison with the wavelength, ray methods and related approximate methods have primary importance in their theory.

Of great practical interest for the elucidation of acoustic focusing systems and for layered sound insulation are problems of the propagation of acoustic plane waves through an aggregate of homogeneous layers or layers with continuously varying parameters. The speaker noted that, although the solutions of such problems are rather cumbersome, at present the theory may be considered well developed both in the case of homogeneous liquid and solid layers. Approximate methods have been developed for calculating the reflection of waves from layers with continuously varying parameters, both in the case of layers thin and of layers thick in comparison with the wavelength.

The report also considered questions of the reflection of bounded beams and pulses and reflection from an uneven surface, as well as questions of the propagation of sounds in statistically inhomogeneous media.

S. M. Rytov investigated the acoustic properties of a finely layered medium. A medium consisting of alternating layers of two homogeneous materials, “on the average,” i.e. for sufficiently thin layers, behaves as homogeneous but anisotropic. The fineness of the layers means that their thicknesses are small in comparison with the wavelengths of compression and shear in their materials.

If the layers alternate periodically, then the problem of wave propagation in such a medium reduces to the solution of a wave equation with periodic coefficients, and the limiting passage gives the solution for a finely layered medium.

In this way the author had earlier solved the problem for the case of electromagnetic waves and had shown that a finely layered medium possesses the properties of a uniaxial crystal. For elastic waves the medium proves to be a crystal of hexagonal symmetry, i.e. it is characterized by five elastic moduli. Expressions have been obtained for these moduli (and for the corresponding velocities of propagation of compression and shear waves in directions along the layers and perpendicular to them) in terms of the densities and elastic constants of the materials of both components. Expressions for the squares of five velocities (two compression waves and three shear waves) coincide with the re-

differ from the results of White and Angona (J. E. White and F. A. Angona, J. ASA 27, 311, 1955), obtained by another method: by considering the static deformation of a cube cut from a finely layered medium.

In the limiting case of thin layers, the periodicity of their alternation is no longer essential, i.e., the layer thicknesses may be varied in an arbitrary manner, but, of course, with the condition that they remain small. Only the relative thicknesses of the two materials enter into the final formulas.

In addition to the case of two solid materials, the case of alternating solid and liquid layers is also considered. The character of the anisotropy here is the same, but for compression waves traveling along the layers, there turn out to be two propagation velocities, while shear waves with displacements across the layers when propagating along the layers, and with displacements along the layers when propagating perpendicular to them, have zero velocity.

This result is valid in the first approximation, when the liquid is regarded as ideal.

Taking account of losses in the solid materials and of the viscosity of the liquid makes it possible, along with the propagation velocities, to obtain the corresponding absorption coefficients for waves of different types.

Artificial anisotropic materials, whose manufacture is not generally difficult, may possess extremely diverse properties and the possibilities of their application are at present still hard to foresee in full. One of the applications may be vibration isolation over a wide range of frequencies. The theory set forth makes it possible to predict the properties of artificial anisotropic materials.

At the meeting A. V. Rimskii-Korsakov presented the report “Investigation of the audibility of nonlinear distortions in the transmission of music and speech by an electroacoustic path,” containing a survey of investigations of the audibility of nonlinear distortions; the investigations were carried out at the Department of Radio Broadcasting and Acoustics of the Leningrad Electrotechnical Institute of Communications named after Bonch-Bruevich in 1953–1955 under the author’s direction. Apparatus was developed that made it possible to introduce various types of nonlinearities in prescribed doses into an electroacoustic sound-reproducing path. The thresholds of noticeability and intolerability of distortions were measured, and mean values and probable deviations were determined for sound material of different character. This made it possible to compile approximate norms of permissible distortions for various classes of electroacoustic transmission.

Having made certain assumptions concerning the character of the fundamental sound signal, the speaker calculated the probability that the amplitude of the nonlinear distortions would exceed the masking threshold (determined by the fundamental signal and by the noise level of the auditorium). In doing so he assumed that the signal is a sum of oscillations with random amplitudes and phases, whose probability density obeys the normal distribution law. These and certain other assumptions allowed the author to calculate the probability of audibility of the quadratic distortions of a signal whose energy frequency spectrum has a constant spectral density within the range 100–5100 cps. It turned out that with a 5% clipping factor, 15–19% of the entire transmission may be spoiled by distortions noticeable to the ear.

Further, the report described apparatus for investigating the laws of distribution of the instantaneous values of a radio-broadcasting signal and presented some measurement data obtained, confirming the validity of the assumptions underlying the calculation.

A. F. Veklenko spoke on “Investigation of the audibility of distortions of the mutual-modulation type.” The general picture of distortions accom-

MEETINGS AND CONFERENCES

systems reproducing sound material (speech, music), is very complex. In order to simplify the phenomena and to subject individual details separately to study, sound-reproducing systems are usually tested by one or several sinusoidal signals. Studies of this kind may, in particular, serve for predicting the evaluation of a sound-reproducing system by listeners (when ordinary sound material is reproduced with its aid).

Of special interest is the study of the audibility of particular varieties of nonlinear distortions of the type of mutual modulation, arising when systems are tested by signals of simplified form. The author used the comparison method: observers were asked to answer whether any difference was audible between two successively presented signals. By presenting for comparison a pure tone and a modulated tone, or two tones modulated to different depths, it was possible to determine the differential thresholds of discrimination of distortions of this kind. No clearly expressed frequency dependence of the audibility of distortions was noted when the frequency of the lower tone was varied within the range from 20 to 200 c/s and the frequency of the upper tone within the range 1000–5000 c/s. The magnitudes of the threshold of distinction of distortions do not depend markedly on the occupied spectrum and on the absolute level of the distortions within the limits from 0 to 25% of the coefficient of mutual modulation. When listening under the conditions of an ordinary cinema-theater hall, observers note a change in the magnitude of the distortions by 2–3%. Since the tests were carried out with a small number of observers, the results obtained may be regarded only as preliminary and require clarification. The speaker considers that it is expedient to evaluate the levels of various kinds of distortions not by means of numerical values of the corresponding “distortion coefficients,” but by the level of their differential thresholds.

The report by V. M. Volf, “Nonlinear Transformation of Oscillations of Complex Form,” was devoted to the development of a method for measuring nonlinearity with the aid of pulses of various forms. It is known that periodic pulses are used to study frequency and phase distortions in linear four-terminal networks. Extending the pulse-measurement method to the measurement of nonlinearity of a four-terminal network, the author modified the block diagram of the dynamic-spectrum method (Radiotekhnika, No. 2, 1953), proposed by him earlier. The essence of the method amounts to the following: from the spectrum of the oscillation arriving at the input of the four-terminal network under test, a certain frequency band is removed with the aid of a narrow-band rejection filter. At the output of the four-terminal network, by means of a suitably tuned band-pass filter, the voltage is measured in the band of frequencies that were suppressed by the rejection filter at the input.

The greater the nonlinearity of the four-terminal network, the greater the product of nonlinearity falling into the frequency band passed by the band-pass filter. By retuning the rejection and band-pass filters over the frequency range, one can measure the nonlinearity product and the signal in a specified portion of the range. The measure of nonlinearity in this case is taken to be the ratio of the measured values of the nonlinearity product to the signal.

In this connection it is borne in mind that the rejection filter practically does not distort the pulse shape. The products of nonlinearity were investigated in nonlinear systems of the form \(i = a_1 u(t) + a_2 u^2(t)\) and for an input signal representing a periodic sequence of triangular, sawtooth, rectangular, exponential, and bell-shaped pulses of equal height, symmetrical with respect to the time axis.

The speaker found that:

a) Other conditions being equal, the nonlinearity product formed in both nonlinear systems depends on the form of the signal. The smallest

the nonlinearity product is formed with a triangular and sawtooth signal, and the largest—with a rectangular signal.

b) If in both systems under consideration the harmonic coefficients are equal, then for any of the signals considered the nonlinearity product in the system with a cubic term will be 2.2–2.5 times greater than the nonlinearity product in the system with a quadratic term. As applied to sound-transmitting systems, this may serve as an explanation of the experimentally observed fact that the effect on hearing of nonlinear distortions of speech, music, etc. in a system with a cubic term is more acute than in a system with a quadratic term.

c) In the case of tests by means of rectangular pulses, linear distortions in the four-terminal network being tested (especially in the circuits following the nonlinear element) lead to a smaller error in the assessment of nonlinearity than when measurements are made by other methods.

d) When rectangular pulses are used, sufficient reliability of measurements can be obtained with a fixed filter setting, i.e., when measuring in one section of the frequency range. This considerably simplifies the measuring device and the measurement procedure itself.

Thus, the method of investigating four-terminal networks by means of rectangular pulses gains generality, becoming suitable for evaluating not only linear properties but also for measuring the nonlinearity of four-terminal networks.

B. D. Tartakovsky reported on the sound reinforcement of open spaces by distributed loudspeaker systems. In such systems (distributed loudspeaker systems—DLS), small loudspeakers are arranged according to a certain plan so densely that the sound of individual loudspeakers ceases to be noticeable and the sensation of a “sounding space” arises. The loudness level at the listening positions becomes more uniform, acoustic feedback between the loudspeakers and the microphone is reduced, and it becomes possible to suppress to some extent disturbances of the echo type. These and other advantages of DLS prompted experimental studies whose purpose was to determine the expediency of using it for sound reinforcement of open spaces, in particular the territory of the All-Union Agricultural Exhibition.

In 1951 the sound-pressure field of a DLS, made in the form of a chain of loudspeakers mounted on masts, was investigated. In this case the distances between the masts and the suspension height varied. The near and far zones were investigated for reproduction of octave bands of continuous noise, music, and speech. It was found that at distances exceeding half the pitch of the chain, the nonuniformity of the loudness level at medium frequencies does not exceed 3 dB and is practically not noticed by the listener. This result is almost independent of the absolute values of the suspension height and the pitch of the chain.

An invited group of specialists compared, in the course of listening tests, the sound quality of various variants of DLS and of dispersed and centralized sound-reinforcement systems. With the exception of two persons, all subjects recognized the advantages of DLS and, in particular, found that the best is a DLS characterized by a distance of 15–20 m between masts at a suspension height of 5 m. Optimal loudness levels were also found for speech and music reproduction at different levels of noise interference. It turned out that the majority prefer a comparatively small amplification of musical transmission, as well as of speech.

The centralized sound-reinforcement systems now in use, because of considerable nonuniformity of the sound-pressure level, cannot provide such an optimal transmission level over most of the area being sounded, whereas DLS makes it possible, within wide limits, to control the distribution of sound-pressure levels over the area being sounded.

and sharply weaken the possibility of echo formation from the surrounding buildings. As a result of the investigations, the advisability was confirmed of the author’s proposal to use distributed loudspeaker systems for sound reinforcement of the territory of the All-Union Agricultural Exhibition. Such a system was designed (chief engineer of the project S. I. Grachev) and built (chief engineer I. N. Shamin) in 1952–1953.

The distributed system of loudspeakers installed on the territory of the All-Union Agricultural Exhibition contains more than 250 radial loudspeakers of the RGD-25 type with a circular directional characteristic, specially manufactured by industry for sound reinforcement of the Exhibition. They are arranged in the form of one or several chains along alleys and ring-shaped chains on the main squares.

The RGD-25 loudspeakers are mounted together with lamps on masts and in lighting fixtures, or are arranged in the form of torchères combined with lamp torchères. In the course of the experiments, certain interesting properties of sound sources were discovered, e.g., the formation of imaginary sound sources, which were subsequently subjected to special investigation.

Comparison of the experimental data with theory made it possible to establish the limits of applicability of the energy theory of distributed systems developed by L. D. Rozenberg, and also to test several simple methods, proposed by the author, for calculating distributed systems intended for the sound reinforcement of open spaces.

In the report by M. I. Karnovskii, “Frequency Characteristics of Some Distributed Systems of Coherent Radiators,” the mutual influences of loudspeakers during the operation of distributed public-address systems were investigated.

In calculations of distributed sound-reinforcement systems, the interaction of loudspeakers, which leads to a change in the radiation resistance of individual loudspeakers, has until now not been taken into account. However, when coherent radiators are placed at distances small or comparable with the wavelength, the radiation resistance of a radiator in a group may differ substantially from the radiation resistance of an isolated radiator. Therefore, despite the fact that single loudspeakers of small dimensions radiate low frequencies poorly, when such loudspeakers are placed sufficiently close to one another in a distributed system it is possible to obtain good output at low frequencies. In this case the distance between radiators should be chosen on the basis of the necessary amount of correction of the frequency characteristic at low frequencies.

The speaker calculated the values of the active component of the radiation resistance of certain groups of primitive spherical radiators of different orders and of groups of complex spherical radiators for the radiation of harmonic oscillations, as well as signals stationary in the probabilistic sense and characterized by an autocorrelation coefficient. Numerical values of possible resistances were calculated for dipoles, radiators of zero plus first order, so-called acoustic columns, radiators forming a plane square lattice, etc.

The report by M. V. Laufer, “Investigation of Methods for Measuring the Nonuniformity of Motion of Sound Carriers,” concerned two new methods for measuring the frequency and the coefficient of nonuniformity of carrier motion in signal recording. In the photographic method of recording, the method of superposing phonograms is suitable. A harmonic signal of a given frequency is recorded on the apparatus under investigation; the negative of the phonogram is cut into two parts, and then the phonograms are superposed on one another so that their teeth coincide. When the motion of the carrier is nonuniform, the phase of the recorded oscillation changes continuously and periodically according to the law

changes in speed. When the superposed phonograms are viewed through a magnifying device by transmitted light, the teeth will periodically converge and diverge. From the magnitude of the maximum divergence of the teeth and the distance between the points of coincidence, the nonuniformity coefficient and the frequency of the speed oscillations of the carrier can be calculated.

According to another “phase-detector” method, a periodic (for example, harmonic) signal of a specified frequency is recorded on the apparatus under investigation by a photographic or magnetic method. The phonogram obtained is reproduced by means of two reading elements placed side by side. The signal reproduced by one reading element lags in time behind the signal reproduced by the other reading element by an amount equal to the ratio of the distance between the reading elements to the nominal speed of the carrier. The signals arising in the sound pickups with a time shift, after preliminary amplification in two channels, are limited in order to eliminate amplitude modulation and are fed to the input of a phase detector. At the output of the phase detector, after a low-pass filter, an alternating voltage appears, which is the envelope of the beats of the input signals. The amplitude of the output voltage is proportional to the nonuniformity coefficient, and its frequency is the frequency of the speed oscillations. After the low-pass filter the signal passes through a system of narrow-band filters and is measured by a vacuum voltmeter. In this way, the frequencies of the speed oscillations and the nonuniformity coefficients are measured for any law of speed oscillation, for each harmonic component separately. In the magnetic recording method, a double reproducing magnetic head with a gap spacing on the order of 8–10 mm may be used; it is installed in place of the recording or reproducing head.

Both methods make it possible to measure a nonuniformity coefficient of the order of 0.01% with an accuracy of up to 3–5%.

R. G. Ofengenden reported on magnetic recording of pulses. The principal problems in the magnetic recording of pulses in the “memory units” of electronic computing machines are the recording of the maximum number of pulses per unit length of the carrier and the recording of pulses at high following frequencies. The maximum permissible number of pulses recorded per unit length of the carrier is called the resolving power of the storage device. The dependence of the resolving power on individual parameters of the path in contact and contactless recording was investigated, and it was shown that with an increase in the magnetizing force the resolving power decreases and, at the same time, the amplitude of the output voltage increases. The optimum value of the gap of the head in contact recording is about 10 μ. The dependence of the magnetic flux in the reproducing head and of the output voltage on the gap between the carrier and the head during the recording of single pulses was also investigated, and it was shown that as the distance between the carrier and the head increases, the resolving power and the amplitude of the output voltage decrease. To determine the influence of losses due to eddy currents on the resolving power, reproduced pulses were recorded at various carrier speeds. When the carrier speed was changed from 2 to 23 m/sec, no noticeable increase in the length of the reproduced pulses was observed. The investigations were carried out with ferrite heads and heads made of permalloy plates.

The speaker then gave the results of a theoretical calculation of the magnetic flux in the reproducing head during the recording of single pulses, showing that the magnetic flux in the reproducing head from the longitudinal component of the magnetization intensity is equal to the flux from the transverse component of the magnetization intensity during the recording

of single pulses. The relations have been studied in detail for cases when the distance between the carrier and the head is greater than the width of the front edge of the target.

An experimental installation was built, on which pulses were recorded at repetition frequencies up to 130 kc/s. The experimental data obtained in 1950–1952 agree, in the main, with the theoretical results.

In the communication by A. G. Almukhamedov, “The Pulse Method for Measuring the Nonuniformity of Sound-Carrier Motion,” the possibility was discussed of measuring fluctuations in the speed of motion of a carrier by recording, on an electronic oscilloscope, oscillations of the time delay between pulses recorded on magnetic tape and the corresponding pulses reproduced from it. The delay time is inversely proportional to the speed of the sound carrier.

In the report “Certain Problems of Modern Magnetic-Recording Technology,” G. S. Veksler noted that magnetic recording has almost completely displaced all other types of recording from radio broadcasting, is being successfully introduced in cinema, and has found application as a “magnetic-memory” unit in calculating and computing devices, etc. In 1954, recording on magnetic tape of television broadcasts was demonstrated. Modern magnetic-recording equipment makes it possible, at a speed of 17.5 mm/sec, to provide speech transmission (up to 2.5 kc/s), and at a speed of 95 mm/sec—to transmit music (up to 10 kc/s) with a dynamic range greater than 65 db and with a nonlinear-distortion coefficient of the order of 2%. One of the essential links of the magnetic-recording path is the carrier—tape or wire. Previously, only wires of carbon steel were used ($B_r = 8000$ G, $H_c = 20\text{–}30$ Oe), and then powder tapes (for type-c tapes, $B_r = 400\text{–}500$ G, $H_c = 100\text{–}180$ Oe). The small value of $H_c$ in comparison with $B_r$ for wires led to considerable self-demagnetization of the carrier at high frequencies. The drop at a frequency of 10 kc/s, even at a speed of 770 mm/sec, amounted, for this reason alone, to more than 20 db. It was established that the output at low frequencies is determined by the value of $B_r$, while at high frequencies it is determined by the value of $H_c$. But although this is true for wires, it is quite untrue for powder carriers—tapes and wire carriers made of stainless steel, for which $H_c$ is commensurate with $B_r$. The erroneous notion that the output at high frequencies for modern carriers depends mainly on $H_c$ led to the search for highly coercive tapes with a large ratio

$$ \frac{H_c}{B_r}. $$

Analysis shows, however, that the ratio

$$ \frac{H_c}{B_r} = 0.2 \div 0.4 $$

is quite sufficient, and that any further increase in it gives an increase of output at high frequencies of less than 2 db. With an overall fall-off of the high frequencies in the path of about 17 db, such a rise plays no significant role. It makes sense to make highly coercive tapes only with a simultaneous increase in the value of $B_r$, so that

$$ \frac{H_c}{B_r} $$

remains within the limits $0.2 \div 0.4$. This will make it possible to obtain an increase in output at all frequencies while preserving the former frequency characteristic.

The speaker then turned to questions of quality control of magnetic-recording carriers. Up to the present, control has been carried out according to electroacoustic parameters, which requires complex apparatus and, because of its bulkiness, makes it possible to test only a small fraction of the production manufactured. It is much simpler and more convenient to control the carrier according to magnetic parameters (for powders this is the only type of control); however, it has still not been possible to establish fully

MEETINGS AND CONFERENCES

a complex mutual dependence between the electroacoustic and magnetic properties of the carrier.

The measurement of the magnetic parameters of carriers can be brought closer to the actual conditions of their operation by carrying out investigations on an apparatus that makes it possible to take into account the character of the field at the recording head, the signal amplitude, and the ultrasonic displacement. The methodology and apparatus for such studies were reported in 1951.

S. G. Gershman reported on an instrument for measuring noise-correlation coefficients, developed by her jointly with E. L. Fainberg.

In connection with the wide use of correlation functions in the analysis of noise, there arises the need for an experimental determination of the parameters used by the theory. One of the quantities that is essential from this point of view is the correlation coefficient.

The instrument developed—a correlometer—is based on measuring an output effect that is proportional to the probability of coincidence of the signs of the instantaneous values of two noises whose correlation is being measured. The electrical circuit of the instrument uses the conversion of the input voltages into rectangular pulses. The converters are two identical amplifier channels with symmetrical limiting in each stage. The output part of the circuit is controlled by an electronic relay that operates only for one predetermined combination of the signs of the input rectangular pulses, which have the same sign as the input voltages. The pulses at the output of the electronic relay, rectangular in form, have a duration no greater than in each of the channels. After averaging the pulses, a constant output current proportional to the probability of coincidence of signs at the output is measured by a pointer indicator or by an automatic recorder.

The theory of the operation of the instrument gives a relation between the probability of coincidence of the signs of the input voltages and the correlation coefficient for noises possessing normal correlation. The results of the theoretical analysis were used for calibration and for estimating the measurement errors. It was shown that the instrument can be used for a number of tasks, for example, for measuring autocorrelation and mutual correlation of noises, for detecting a weak signal against a noise background, and for various measurements in a sound field. The report gave an example of experimental use of the instrument—autocorrelation curves (records of the mutual correlation of two voltages), which are, in turn, sums of two noises (in one case, for sound oscillations at two points of an enclosed room; in another, for oscillations at one point of the room and a voltage supplied to a loudspeaker). S. G. Gershman also gave an estimate of the diffuseness of the sound field in the room and indicated that the possible applications of the correlometer are not exhausted by the instruments considered.

N. F. Vollernér touched upon questions of instrumental spectral analysis. The results of analytical and experimental determination of the frequency spectrum of the process \(E(t)\) (\(t\) is time) are not identical. This is explained by the fact that the analytical frequency spectrum \(E(\omega)\) (\(\omega\) is the angular frequency) is a functional of \(E(t)\) (depends on all values of \(E(t)\)). The readings of frequency analyzers—spectrometers—at each given moment \(t_p\) are determined by the values of \(E(t)\) only in the interval from \(t_1-\Delta t\) to \(t\) (where \(\Delta t=\frac{1}{\Delta f_{\text{an}}}\), and \(\Delta f_{\text{an}}\) is the pass band of the analyzer). Spectral analysis of physical processes performed with the aid of various analyzers was proposed by the author to be called instrumental spectral analysis. As a result of instrumental analysis, an instantaneous frequency spectrum \(u(t,\omega)\) is obtained—a sequence of levels, maximal over a time of order \(\Delta t\), at the output of a set of filters to which the voltage under study is applied.

At identical energy levels, the spectra, depending on the character of the process (fluctuation, pulse, etc.), will have different instantaneous frequency spectra; i.e., the result of instrumental spectral analysis is determined not only by the level of the process under study, but also by its character. For a complete representation of the process under study, one needs a set of instantaneous frequency spectra for different moments of time during the course of the process. In view of the spectrograms of the process \(E(t)\), it is not difficult to determine the character of the process: for fluctuational voltages the character of the envelope of the spectrogram does not change, but the readings of the spectrometer at the given frequency (with a short averaging time) change noticeably all the time; for periodic processes the spectrograms are more or less stable, etc.

For practical applications there is no need to reduce the results of instrumental spectral analysis to the concepts of an analytical frequency spectrum, since real instruments operating from the phenomenon being analyzed have a finite pass band \(\Delta f_{\text{instr}}\). Consequently, their output level at a given moment is determined by the entire process, and also, as in instrumental spectral analysis, by the part of the process in the time interval of order \(1/\Delta f_{\text{instr}}\).

If the instantaneous frequency spectrum \(u(t,\omega)\) is known, the output voltage of the instrument for the fluctuational character of the process under study is equal to

\[ u_{\text{pr}} = u(t,\omega)\sqrt{\frac{\Delta f_{\text{prib}}}{\Delta f_{\text{an}}}}, \]

and for the pulse character

\[ u_{\text{pr}} = u(t,\omega)\frac{\Delta f_{\text{prib}}}{\Delta f_{\text{an}}}. \]

The speaker considers that spectrum analyzers should be designed taking into account the parameters of those instruments upon which the analyzed processes will act, and in their construction the results of instrumental spectral analysis should be used. In the range of infrasonic frequencies and low audio frequencies it is advisable to make spectrometers with a simultaneous method of analysis (filter type), and in the audio and ultrasonic range—with a sequential method of analysis (heterodyne type). Volderner reported that at the Department of Radio Receiving Devices of the Kiev Order of Lenin Polytechnic Institute a set of spectrometers had been developed for the range \(2\ \text{cps}—500\ \text{kcps}\), consisting of four instruments: \(2—500\ \text{cps}\)—filter type; \(0.4—10\ \text{kcps}\); \(3—100\ \text{kcps}\); \(50—500\ \text{kcps}\)—heterodyne type.

The report by M. S. Antsyferov contained calculation data for a vibrometer with spring suspension developed by him. The author commented in detail on the calculation of the frequency characteristic of the instrument at frequencies lying above the region of the natural frequency. For the region of low frequencies, within which all the principal elements of the instrument can be characterized by lumped mechanical parameters, the frequency characteristic is obtained as rectilinear.

In passing into the region of high frequencies, the distributed mass and elasticity of the main springs on which the moving part of the vibrometer is mounted begin to have an effect. The author considered, as a particular example, an inertial vibrometer having a suspension of the inertial mass on flat springs with so-called semi-clamped ends, the influence of the distributed parameters of the main mounting springs on the frequency characteristic of the vibrometer. He regarded the remaining parameters of the vibrometer as lumped. The forced vibrations of the spring, excited at one end, were calculated.

with a prescribed displacement. At the other end a load of arbitrary magnitude and an active resistance are assumed, which does not bring the system, in the range of frequencies under consideration, into an aperiodic regime.

The author obtained formulas for the frequencies of secondary resonances and the antiresonances accompanying them. Secondary resonances and antiresonances can be damped by placing on the surface of the spring a viscoelastic material (rubber, Viscolloid). By using specially profiled springs with elastic ends and a rigid middle part, secondary resonances and antiresonances can be shifted outside the operating range.

B. G. Belkin spoke about a new generator for acoustic measurements and about experience in its application. In this generator a “sliding” noise band of any relative width is produced, capable of continuously moving over the range of sound frequencies. For this purpose the noise band is recorded on motion-picture film and reproduced at a variable speed. The audio-frequency path of the generator consists of a loop of film with the recorded phonogram, sound optics, a photocell, and a low-frequency amplifier—standard elements of a cinema sound-reproducing system.

With the aid of a special mechanical device, the speed of motion of the phonogram is periodically changed by a factor of 100 according to a logarithmic law. If, for example, a band of continuous noise of \(1000 \pm 50\) cps is recorded on the phonogram at normal speed, the mean frequency of 1000 cps, when the speed of rotation of the drum is changed, will vary from 100 to 10,000 cps. In this case the relative width of the reproduced frequency band will remain constant, always amounting to 10% of the mean frequency. A change in the relative width of the noise band is achieved by replacing the phonogram. The rate of frequency change is easily varied over wide limits and can be synchronized with the speed of movement of the tape in a self-recording measuring instrument.

The author approximately calculated the magnitude of detonation permissible from the standpoint of broadening of the spectrum of the noise band. He found, for example, that for a relative broadening of the noise band due to detonation equal to 0.4, with a relative width of the noise band recorded on the phonogram amounting to 0.05, and with the spectral density of the noise band at the limiting frequencies, in fractions of the maximum spectral density, equal to 0.01, the permissible detonation is less than 1.5%, which is readily attainable.

The author then described some cases of application of the generator he had developed.

V. M. Gardash’yan spoke about an instrument for measuring certain acoustic parameters of rooms, developed by him jointly with A. N. Kacherovich. For acoustic measurements in rooms, at present the fast-acting level recorder of the “Neiman” type and various electrical apparatus are used as recording instruments. At the same time, in recent years the country has been carrying out a mass reconstruction of theaters, cinemas, and clubs characterized by unsatisfactory acoustic conditions. Therefore there arose the need to create simple and sufficiently accurate acoustic measuring equipment for investigating reconstructed rooms and accumulating statistical data on room acoustics.

The instrument developed and manufactured is intended for recording curves of the decay of sound energy over a wide frequency range and, in prescribed frequency bands, for determining the reverberation time and for recording the sound energy in the steady-state regime. It consists of a microphone, a preamplifier, a logarithmic amplifier stage with a detector, and a recording device.

The instrument has a set of filters that make it possible to separate out five octaves (three in the low-frequency range, one in the middle-frequency range, and one in the high-frequency range). The logarithmic stage makes it possible to record processes within a range of 50 dB with an error of ±1 dB. The instrument operates stably regardless of the scatter of tube parameters and fluctuations of the supply voltage within ±15%. The inertia of the entire circuit makes it possible to record decay curves beginning with \(T = 0.4\) sec. The recording speed is not less than 300 dB/sec. The frequency range of the instrument is 75–16,000 Hz, with an overall nonuniformity of ±2 dB. When operating with a microphone, the frequency characteristic is determined mainly by the characteristic of the microphone. The recorder of the instrument writes in ink on a diagram paper strip with a scale width of 60 cm (a pen deflection of 1 mm corresponds to 1 dB). The instrument provides for the introduction of artificial inertia. The paper speed is 10 mm/sec. The weight of the instrument is about 20 kg.

I. E. Elpiner spoke about the biological action of waves.

Ultrasonic waves cause not only mechanical ruptures of cells and cellular structures, as had previously been assumed. More subtle, partially reversible changes in certain cell functions are possible without microscopically detectable injury to the cell. It has turned out that, under the influence of ultrasound, intracellular molecular complexes break down or “are shaken loose,” accompanied by inhibition of some or enhancement of other enzyme systems—biocatalysts.

For example, the speaker established experimentally that in yeast cells the enzyme that decomposes sucrose becomes many times more active if these cells have first been subjected to the action of ultrasound. Moreover, under the action of ultrasound this enzyme is found in yeast cells in which it is normally absent. The enzyme cholinesterase, found in brain tissue and associated with the process of transmission of nervous excitation, increases its activity under the action of ultrasound. In the speaker’s laboratory, data have been obtained showing that in sonicated cells the quantity of other biologically active substances—certain vitamins—also increases. All this leads to the idea that ultrasound, by destroying molecular complexes, releases biologically active substances that play a significant role in the vital activity of the cell and the organism.

To date, the appearance of cavitation in the protoplasm of a cell under the action of ultrasound has not been proved. It is possible that the phenomena described above are due to the fact that, under the action of ultrasound, the colloidal structure of cellular elements is disrupted (acceleration and orientation of particles, etc., play a role here).

I. E. Elpiner emphasized that the mechanism of the action of ultrasound on intracellular molecular complexes is subject to comprehensive study.

B. D. Tartakovsky