Volume 28, Issue 5, September 1956
Index of content:
28(1956); http://dx.doi.org/10.1121/1.1908487View Description Hide Description
Charts and formulas are presented for the calculation of the loudness of noises having approximately continuous spectra. By means of direct loudness matches it is shown that the loudnesses in octave bands can be combined according to the formula , where St is the loudness (in sones) of the total noise,Sm is the loudness of the loudest band, and ΣS is the sum of the loudnesses of all the bands. The loudnesses of the octave bands can be determined from measurements of their sound pressure levels by means of a chart based upon a new determination of the equal loudness contours for bands of noise. Other charts and formulas are presented for half‐ and third‐octave band widths.
Measurements were also made of the dependence of loudness on such factors as (a) separation between noncontiguous bands, (b) width of very narrow bands, and (c) rate and level of square wave modulation.
28(1956); http://dx.doi.org/10.1121/1.1908489View Description Hide Description
In order to discover what are the maximum noise levels that office personnel find acceptable and what their reactions are to noisy offices, a survey was carried out at a large air base. A questionnaire composed of 15 rating scales was administered to 190 people scattered over 17 different locations on the base. The rating scales allowed the workers to assess such things as the “noisiness” of their environment and to appraise the effect of noise on various aspects of their work, such as their ability to converse or to use the telephone.
The results obtained with the rating scales were compared with various physical measures of the noise. A high correlation was found between perceived noisiness and the measure called “speech interference level” SIL, which is the average of the sound pressure levels measured in the three octave bands between 600 and 4800 cycles per second. An even higher correlation was found between the ratings of noisiness and a computed measure of loudness level. More than two‐thirds of those questioned stated that speech communication was an essential part of their activities and that the more intense noises in their offices interfered with it. These correlations provide a basis for setting criteria for the maximum noise acceptable in terms of SIL and of loudness level. These results, plus those from a previous study, suggest that the maximum continuous noise levels acceptable for office spaces in which speech communication is important should not exceed an SIL of 40 db. Office employees are accustomed to a noise spectrum of a form that yields a loudness level of about 22 units higher than the SIL. Complaints are encountered when the loudness level exceeds the SIL by 30 or more units. Thus, with an SIL of 40 db the loudness level should preferably not exceed about 62 phons. Only under special conditions with this SIL would it be advisable to exceed a loudness level of 70 phons.
For intermittent noises, such as the noises produced by aircraft operations, evidence is presented to show that the mean (average) value of the SIL and loudness level should not exceed the numbers for steady noises if the work of office employees is not to be disturbed.
28(1956); http://dx.doi.org/10.1121/1.1908491View Description Hide Description
The threshold of overload and the growth in intensity of the aural harmonics are determined by the exploring tone method for frequencies of 100, 350, 1000, 2000, and 5000 cps. It is found that the level of a test tone at which it first overloads the ear can be quite accurately determined and that the range of beats between the second harmonic and the exploring tone, as the latter is changed in intensity, is considerably narrower at the overload threshold level than at higher levels. As the test tone is raised in intensity the loudness of the harmonic increases rapidly.
Overloading occurs at approximately the same sound pressure level for all frequencies so that the range of linear response is around 8 db at 100 cps, 30 db at 350 cps, and 50 db at 1000, 2000, and 5000 cps. All harmonics are detected when they reach a certain consistent level above threshold for that frequency. The exact level depends upon the subject.
These results are considered as additional evidence that distortion in the ear takes place within the sensory cells as part of the electromechanical process and is not due to any nonlinearity of the mechanical response of the ear.
28(1956); http://dx.doi.org/10.1121/1.1908493View Description Hide Description
Thresholds for the detection of interaural time difference were determined by ten listeners (1) for band‐limited random noise (150–1700 cps), (2) for a 1000‐cps tone, and (3) for a 1‐millisecond click. The average interaural time differences corresponding to 75% correct detection in the symmetrical two‐alternative tests were (1) 9 microseconds, (2) 11 microseconds, and (3) 28 microseconds. Ranges of individual thresholds and group psychometric functions are presented.
28(1956); http://dx.doi.org/10.1121/1.1908495View Description Hide Description
The just noticeable difference in dichotic phase, as a function of sensation level and of frequency, has been determined on a number of listeners with normal hearing. The test tones were transmitted by earphones, and the phase difference between the ears was varied by means of an electronic phase shifter. The psychophysical method used combined paired comparisons and forced choice. The first tone pulse of each pair presented was kept at a constant phase difference at which the subject localized the sound source as equidistant from his ears. The dichotic phase difference of the second pulse was varied irregularly (“randomly”). The results show that the sensitivity to dichotic phase difference is highest (2° of phase) at medium sensation levels, and that the jnd increases with positive acceleration as the sound frequency increases. Around 1300 cps the jnd becomes so great that it cannot be measured. The dichotic time difference calculated from the measured jnd in phase has a minimum near 800 cps.
28(1956); http://dx.doi.org/10.1121/1.1908497View Description Hide Description
Loudness adaptation may be measured by a simultaneous loudness balance, or, as in the two experiments reported below, by a median plane localization of a dichotically presented acoustic stimulus. The loudness of a steady auditory stimulus generally decreases with time. That is, the intensity of a comparison stimulus in the rested ear is ordinarily set below the intensity in the stimulated ear.
Two experiments were done on loudness adaptation for bands of noise. In Experiment 1, using 36 subjects, loudness adaptation for a wide‐band thermal noise of 100–5000 cps was studied as a function of five SPL's: 40, 70, 90, 100, and 105 db over all. The mean maximum loudness adaptation obtained was 2.3, 9.9, 11.4, 14.4, and 16.3 db, respectively. The mean standard deviation for all measures was 6.1 db, and the distributions of the sets of measures tended to be skewed toward greater adaptation.
In Experiment 2 (12 subjects), the loudness adaptation for a 1500‐cps tone was compared with that for bands of noise whose centers (mel scale) were at 1500 cps, and whose over‐all SPL's were equal to the SPL of the pure tone. The band limits in cycles per second were 1280–1120, 1075–1950, 720–2600, and 100–4900. For each band, adaptation was measured for 50, 70, and 90 db SPL over all. The essential results are: (1) Loudness adaptation for 1500 cps is about 8.5 db greater than the maximum adaptation for any noise band at any SPL; (2) Adaptation is small (4.5 db) at 50 db for all bands of noise; it is complete within one minute and is about equal for all band widths; (3) At 70 and 90 db, time taken for complete adaptation increases and the two widest bands give greater adaptation than the two narrowest bands. At 90 db a trend becomes clear: the wider the band, the greater the degree of adaptation and the longer the time required for maximum adaptation to be attained.
28(1956); http://dx.doi.org/10.1121/1.1908499View Description Hide Description
Annoyance threshold judgments were obtained by exposing an individual to noise for three minutes and asking him to adjust the intensity to the level which, if any louder, would annoy him if it were present most of the time where he was working.
In one experiment, 21 people made judgments about 13 bands of noise which covered the frequency range of 50 to 13 000 cps, and subsequently made sets of equal loudness matches. No differences were found between annoyance threshold curves and equal loudness curves. In a second experiment, each of 162 people made one annoyance judgment. When these annoyance thresholds were transformed into equivalent loudness terms, the resultant annoyance threshold curve varied reliably with frequency only in that the threshold on the highest band (6600–9000 cps) was reliably lower than those on lower frequency bands.
Office workers who had once worked in noisy situations as well as those working in noisy situations at the time of the experiment gave thresholds about 15 db higher than did people who had only worked in office‐type situations.
Within a group who had worked only in quiet situations, those who tried to imagine themselves in an actual working situation gave thresholds that averaged about 15 db higher than the thresholds of those who did not.
28(1956); http://dx.doi.org/10.1121/1.1908502View Description Hide Description
Tartini pitches and beat tones behave differently and must have different physiological origins. Tartini pitches are heard only if the primary tones remain below 2000 cycles per second, preferably for clear observance below 1000. The cochlea is essential for hearing them. Beat tones are not so limited. Their observation is easiest (least confusing) when the primaries are chosen in the region from 3000 to 5000 cycles per second. But lower tones may be used (although there may then be an overlapping of Tartini pitches and beat tones). The cochlea need not exist for the existence of beat tones.
28(1956); http://dx.doi.org/10.1121/1.1908504View Description Hide Description
The theory of statistical decision has previously been applied to the problem of sensory detection of signals. In this paper, the theory is expanded to treat a simple recognition problem. While the data supporting the expansion have been collected in auditory experiments, the theory applies generally to all human sensory behavior.
28(1956); http://dx.doi.org/10.1121/1.1908506View Description Hide Description
A human listener is here regarded as a “cross‐correlator”; his two ears are treated as “input terminals,” stimulated with the acoustic input signals f 1(t) and , while his vocal responses are treated as the output “correlation function.” The two signals are, respectively, pure and distorted versions of the same signal (perhaps speech). The delay τ is randomly set and the listener answers right or left, as the source of sound appears to him to lie. The “correlation function” then corresponds to the probability distribution of his correct judgments. Such functions represent the degree of aural fusion, and show up strikingly the invariants of speech signals which are significant in aural perception.
28(1956); http://dx.doi.org/10.1121/1.1908508View Description Hide Description
Comparisons are made of the intelligibility of continuous speech in noise under three listening conditions: Speech arriving simultaneously and in phase at the two ears; speech arriving at one ear later than the other; and speech arriving simultaneously, but in opposite phase in one ear. Time intervals between ears ranged from 200 μsec to 7 msec. Delaying speech to one ear under these conditions does increase intelligibility but is little, if any, better than the simple expedient of wiring the phones so that the wave form at one ear is inverted with respect to the other. With the materials and method used, the maximum increase occurs with a delay between ears somewhere between 0.5 and 1 msec, though there is some indication that delays 1 msec and longer may be slightly better under the most difficult listening conditions tried.
28(1956); http://dx.doi.org/10.1121/1.1908510View Description Hide Description
Measurements were made of the intelligibility of speech heard in noise and produced by different amounts of vocal force. Vocal force ranged from the weakest voiced whisper to a very heavy shout. The results show less than 5% deterioration in intelligibility over the range from a moderately low voice to a very loud voice (55 to 78 db in a free field at one m from the lips). Beyond these points intelligibility decreases abruptly and in a linear relation to decibel change in vocal intensity. Listeners' errors are analyzed to determine the effects of the extremes of vocal force on the intelligibility of different parts of the syllable and of different vowels.
28(1956); http://dx.doi.org/10.1121/1.1908512View Description Hide Description
The absolute identification and differential discrimination of the sound level of tones were studied over a wide range of sound levels under comparable experimental conditions, and the results of the two experiments were compared in comparable units. Absolute identification improves and differential discrimination deteriorates as the range of sound levels examined increases. The net effect of these complementary changes is that the information associated with differential discrimination is roughly of the same magnitude as that associated with absolute identification.
28(1956); http://dx.doi.org/10.1121/1.1908514View Description Hide Description
The characteristic feature of the method is that at any point in the sound field the conservation criteria are expressed in terms of the instantaneous values of the speed of propagation, the particle velocity, the excess pressure, and excess density. These criteria, together with the adiabatic assumption, determine explicit relations between any two of these quantities. Excess pressure and excess densities are here defined as departures from the equilibrium values that exist at the instant when the particle velocity is zero.
For waves of finite amplitude these equilibrium values, as well as the speed of propagation, are found to depend on the intensity. The increment in the speed of propagation does not agree with that obtained by classical methods of analysis. The discrepancy found to be due to the omission in the classical forms of the continuity criterion of a term that specifies the effect of the rate of change in the speed of propagation.
28(1956); http://dx.doi.org/10.1121/1.1908516View Description Hide Description
Visualization of the vortexflowpattern in typical jets emitting jet tones was made by means of shadowgraph techniques to show the nature of the ever present process of vortex coalescence in the jet and how this determines the values of the eigenton sound frequencies radiated by the jet. These frequencies are super‐imposed on the noise background radiated by the jet. The dependence of (a) vortex shedding frequency from the orifice, (b) jet‐tone frequencies, and (c) translational velocity of the vortices on Reynolds number (Re) is shown.
These studies were carried out with carbon dioxide jets discharging into the atmosphere. The flow channel geometry consisted of an orifice plate containing a sharp‐edged, circular orifice, 0.250 in. in diameter and 0.093 in. thick attached to a large stilling tank partly filled with sound‐absorbing material to eliminate any effects of cavity resonance on the jet. The studies were carried out over the range from Re zero to Re 7000, where and v is mean velocity of discharge of jet; ρ, gas density; t, orifice plate thickness, and μ, gas viscosity.
28(1956); http://dx.doi.org/10.1121/1.1908518View Description Hide Description
28(1956); http://dx.doi.org/10.1121/1.1908520View Description Hide Description
Measurements of pressure and particle velocity in fluid‐filled bore holes have yielded several unexpected results, some of which can be explained by very simple computations. One is an acoustic square wave, a flat‐topped pulse which is generated when a slug of water, broken off from the main column by cavitation, rejoins the column. In another instance, the geometry is such that a wave along a steel casing is the first signal to arrive at a pressure detector, and the initial pressure is negative, although the explosive pressure generating the transient is initially positive. The manner in which a steel casing causes reversal of pressure in the “casing break” is described. An expression is also presented which relates the negative pressure observed at the bottom of an uncased well to the elastic constants of the fluid and solid. As a fourth example, initial upward motion of a geophone hanging in a bore hole is shown to be consistent with the circumstances under which the observations were made.
28(1956); http://dx.doi.org/10.1121/1.1908522View Description Hide Description
This paper discusses the forbidden and allowed modes of radial vibrations in hollow cylinders of bariumtitanate. Theoretically, for a given ratio of outside diameter to inside diameter, it is possible to predict which of these higher modes will be observed. The theory has been checked experimentally and, within the limits of the experiment, has been found to be true.
28(1956); http://dx.doi.org/10.1121/1.1908524View Description Hide Description
At very low temperatures the mean free paths of electrons in metals become so large with respect to their values at room temperature that they can acquire momentum from collisions with lattice vibrations which result in ultrasonic attenuation when the electrons are in the normal state. This attenuation drops to zero in the superconducting state. Original measurements in impure lead, tin, and copper showed an attenuation proportional to the square of the frequency which could be directly related to the electronviscosity calculated from the number N, the mass m, the mean free path l and the mean velocity 0 of the electrons.
Recently some very pure tin samples have been obtained and much larger effects have been measured due to the increased conductivity at low temperatures. Six oriented samples have been measured and from the measurements the six elastic constants and six viscosity coefficients have been obtained. These correspond to those for a tetragonal crystal. As the mean free path becomes longer than the acoustic wavelength, the loss is determined by a scattering process and the loss for a given frequency approaches a limiting value in agreement with a theoretical prediction of Pippard. The loss for longitudinal waves is from 2.5 to 7.5 times the predicted value while that for shear waves is about 1.5 times the theoretical value of Pippard's.
Measurements of the elastic constants have been made through the super conducting range and the only discontinuity that occurs is that due to the thermodynamically predicted value. The change for a longitudinal wave is less than that predicted for a volume change in agreement with a theoretical derivation. No relaxation effect occurs for the measuredvelocity even for very pure tin.
28(1956); http://dx.doi.org/10.1121/1.1908526View Description Hide Description
The possibilities and limitations of different ultrasonic stroboscopes described by Bär for the special purpose of studying ultrasonic phenomena were investigated in greater detail. One type is appropriate for the measurement of the velocity of propagation of progressive ultrasonicwaves. A modified setup offers special advantages for the precise determination of the sound velocity in a liquid relative to a standard liquid. Another optical arrangement is best suited for the study of ultrasonic fields; it was used for direct measurements of the Rayleigh phase shift at a liquid‐liquid interface and for photography of the phase field before an ultrasonic tranducer.