Volume 24, Issue 1, January 1952
Index of content:
- PROGRAM OF THE FORTY-SECOND MEETING OF THE ACOUSTICAL SOCIETY OF AMERICA
- Session A: Instrumentation
- Contributed Papers
24(1952); http://dx.doi.org/10.1121/1.1917407View Description Hide Description
In the replay apparatus for visible speechfilms, it was first necessary to make the film. Then the film was passed by a slot and the emanating modulated light was inverted into frequencies by the aid of a modulation disk. Phono recordings of music pieces that were first recorded on replay film have a good quality when played back through the replay apparatus at normal, multiplied, and divided modulation disk speed. It was also found possible to pass negative as well as positive picture film patterns through the device. However, a disadvantage of such a system is the time delay and associated work of producing the speechfilm. It will be shown in this paper that it is possible to replay immediately spoken information by using special apparatus in conjunction with the replay device. For this, it is not necessary to pre-analyze the speech band on film. It may be done, for example, by a set of 100 magnetostriction filters, utilizing 30 cps band widths for each filter element for a total band width of 3000 cps. For frequency band analyzing purposes the outputs of the single filter elements are connected to 100 elongated neon tubes. The spectrum distribution is immediately seen by the envelope of the lengths of light portrayed by the 100 adjacent neon tubes. For frequency band compression or expansion, by similarly portraying 100 small neon tubes which are positioned in a small light wave guide, 100 light points are produced along a thin line. Each light point is varied in intensity by one of the elementary filters. Now, instead of a speechfilm, it is possible to project these 100 speech-controlled light points to the modulation disk. Thus, the output may be an instantaneously divided, multiplied, or normal speechspectrum. It is noteworthy that instead of the neon bulb array a smaller special multigrid oscilloscope tube may be used.
24(1952); http://dx.doi.org/10.1121/1.1917408View Description Hide Description
In biological work, rapid measurements with small samples are often necessary. A two-crystal interferometer has been built for liquid or semi-solid samples of 30 to 100 cc. Soundvelocities can be determined to within one part in 500 with a travel of less than 3 mm using a crystal output voltage of about 200 microvolts. Provision has been made for temperature control and for rapid change of crystals to permit measurement at 1, 2, 4, 12, and 36 mc. A phase comparison method has been used for wavelength determinations incorporating a dual heterodyne reduction to audiofrequencies with Lissajous' figures on an oscilloscope. A sonic delay line has been included for absorptionmeasurements using pulsed sound and a calibrated attenuator.
24(1952); http://dx.doi.org/10.1121/1.1917410View Description Hide Description
In a previous paper a method of direct ultrasonography was developed. This paper deals with recent experiments on indirect ultrasonoscopy and ultrasonography mainly using special phosphors. The actual technique involved in the new methods is discussed and their results illustrated by lantern slides.
24(1952); http://dx.doi.org/10.1121/1.1917411View Description Hide Description
The resonant frequency of a crystal transducer may either increase or decrease when taken from air and immersed in water. A formula is derived which predicts the approximate resonance in either case. In general, it is found that the reactive component of the radiation impedance tends to lower the frequency while the resistive component tends to raise the resonant frequency by effectively diminishing the Rayleigh “finite width” correction. The resultant frequency cannot, of course, exceed the “thin bar” resonance.
24(1952); http://dx.doi.org/10.1121/1.1917412View Description Hide Description
The resonant frequency of a crystal and backing plate combination can be controlled by means of a magnetic field applied to the backing plate. A change in frequency of one-half an octave can be attained. The “Q” of the combination remains almost constant with the small changes in the magnetic field associated with the frequency change.
24(1952); http://dx.doi.org/10.1121/1.1917413View Description Hide Description
The interior of forgings as revealed by ultrasonic echo techniques is displayed as orthographic projections on cathode-ray tubes. The scanner,Snell's law computer, and display circuits are discussed and slides are presented showing correlation between the plotted ultrasonicimages and defects in metal.
- Session B: Speech and Hearing
24(1952); http://dx.doi.org/10.1121/1.1917414View Description Hide Description
Work by Hawkins, Hirsh, Licklider, Webster, and others has shown the considerable reduction in masking which occurs when the phase of either the signal or the noise is reversed, or when the phase of the signal is shifted by various amounts. The present paper extends this work by shifting the interaural phases of the masking component of the noise by various amounts between +180° and −180°. The interaural phases of both the noise and the tone are therefore parameters of this study. The results are in agreement with those already published, but show in addition large reductions in thresholds when the noise and tone differ in phase, even when neither has been reversed.
24(1952); http://dx.doi.org/10.1121/1.1917415View Description Hide Description
Articulation tests were conducted under uniform conditions (same talkers, same listeners, randomized block design of experiment) to compare the effects upon intelligibility of various types and amounts of nonlinear distortion. The types included peak clipping (both symmetrical and asymmetrical), center clipping, linear rectification, and parabolic (“square-law”) rectification. The results confirm previous reports that intelligibility is highly resistant to symmetrical peak clipping (80 percent word articulation with infinite peak clipping). They show that asymmetrical peak clipping is almost entirely without detrimental effect upon intelligibility until more than half the speech wave is eliminated (98 percent word articulation with half-wave rectification). In addition, the results provide a basis for comparing effects of odd and even harmonic distortion, for comparing the effects of noise that enters the system at a point ahead of the nonlinear circuit with effects of noise that enters at a point following the nonlinear circuit, and for deciding what are tolerable limits for nonlinear distortion in speech communication systems.
24(1952); http://dx.doi.org/10.1121/1.1917416View Description Hide Description
The comparison of the performance of earphones on the ear and on closed calibrating couplers (reported in November, 1950 and May, 1951) has been extended to a study of the phase relationships involved. From these, the acoustic impedance that the ear presents to an earphone can be deduced. Using a direct recording meter (described by R. F. Brown), the real and reactive components of the pressure developed in the cavity or ear canal were observed. The results of the measurements will be discussed.
24(1952); http://dx.doi.org/10.1121/1.1917417View Description Hide Description
Earlier work by one of the present authors (N.W.) indicated that the equal loudness contours for hearing by bone conduction for normal-hearing listeners might be different in shape from similar contours for hearing by air conduction if both sets of contours were plotted in terms of level-above-threshold, referred to the respective thresholds. A bone-conduction “preferred” equal loudness contour for a listener with almost pure conductive loss had been found to be definitely different from the corresponding air-conduction equal loudness contours for normal-hearing listeners, being more sharply curved when all contours were plotted in terms of level-above-threshold, referred to the appropriate thresholds. Thus equal loudness contours for bone-conduction listening for normal-hearing listeners have been determined, using the electrodynamic bone-conduction vibrator and the air- and bone-conduction audio testing assembly at U.C.L.A., further developed and improved since its first presentation before the Society. Contours were determined for loudnesses corresponding to levels of 20 and 40 db above the bone-conduction threshold at 1000 cycles per second. The listeners' hearing was normal as determined by audiometric and otological examination. The contours, when plotted in terms of level-above-threshold, are definitely more sharply curved than the corresponding air-conduction equal loudness contours, plotted in terms of level-above-threshold (air conduction). A tentative explanation for and the implications of this difference is given for both normal-hearing listeners and those with pure conductive impairment.
24(1952); http://dx.doi.org/10.1121/1.1917418View Description Hide Description
During the course of some experiments on listening by bone conduction, the author noticed an increase in the pitch of a pure tone in the middle audiofrequency region when a pure tone of lower frequency was sounded, and the qualitative results of several ensuing experiments were reported in a letter to the editor. The work to be described here dealt with the rise in pitch of a pure 2000-cps tone in the presence of a broad band of thermal noise and also in the presence of pure tones of lower frequency. All sounds were presented by bone conduction using the air and bone conduction audiotesting assembly developed at U.C.L.A. The results of extensive tests on one observer are presented as well as the results of pitch rise tests on several different listeners. All listeners had normal hearing as indicated by audiograms and otological examinations. It was found that the pitch of the 2000-cps tone rose with increasing level of the disturbing sound and that the greatest pitch rise occurred just before the pure tone was masked. Individual differences were evident among listeners. The effect on the pitch of the 2000-cps tone of clenching the teeth is discussed for one observer. A tentative explanation of the phenomena is given.
The Mechanical Impedance of the Human Mastoid and Its Relation to the Calibration of Bone Conduction Transducers24(1952); http://dx.doi.org/10.1121/1.1917419View Description Hide Description
Knowledge of the compliance and the resistance of the skin overlying the human mastoid is important for the calibration of bone conduction hearing aids. Artificial mastoids for the testing of bone conduction transducers can be successfully designed and standardized only when they actually simulate the mechanical impedance of the human mastoid. The correct value of the impedance, then, can be achieved by a properly chosen “flesh simulating” pad. Because of discrepancies of about one order of magnitude in the measured values of compliance and resistance published so far, this question was taken up again. The impedance of the human mastoid was determined by methods previously developed to investigate the elastic and viscousproperties of human tissue. The impedances of various subjects were measured. The compliance and resistance were calculated from the results and it is shown how they depend on certain parameters. The reasons for the discrepancies mentioned above are explained. The influence on the calibration of bone conduction transducers is discussed.
24(1952); http://dx.doi.org/10.1121/1.1917420View Description Hide Description
Work reported previously on synthetic speech produced from hand-drawn and highly schematized spectrograms has suggested an experimental test of the role of a single acoustic (spectrographic) variable in determining the identifiability of the voiceless stops p, t, and k. Schematized syllables consisting of a stop-like sound followed by a two-formant vowel were presented for identification of the consonant. The schematic stop consisted of a brief burst of tones confined to a narrow frequency range. The center frequency of this burst was varied in 12 equal steps from 360 to 4320 cps. Each schematic stop was paired with each of 7 two-formant cardinal vowels. The results show close agreement between subjects on the choice of certain frequency positions for p, t, and k; a systematic variation of this position from vowel to vowel; and, in certain cases, so marked an interdependence of the stop and vowel components of the schematic syllable that the same schematic stop was judged by a large majority of the subjects to be p when paired with i and u, but to be k when paired with a. Hence, it appears that the syllable rather than the phoneme may, at times, be the irreducible acoustic stimulus.
- Session C: General Acoustics
24(1952); http://dx.doi.org/10.1121/1.1917421View Description Hide Description
The master control system installed in Montreal headquarters of the Canadian Broadcasting Corporation, apart from being one of the largest on the North American Continent, is noteworthy for the fact that all the system wiring is carried out in unshielded telephone cable. It is estimated that nearly half a million feet of wire were used. In spite of this, the signal/crosstalk ratio has been kept better than 75 db for interfering signals (at full level) of frequencies up to about 4000 cycles and better than the RTMA specification figure of 65 db up to about 10,000 cycles. This paper describes the methods and precautions used throughout the installation to produce these results.
24(1952); http://dx.doi.org/10.1121/1.1917422View Description Hide Description
The phangle meter is an instrument for recording automatically the in-phase and quadrature components of the output of a four-terminal network with reference to its input. The present instrument operates over a range of 50 cps to 10,000 cps and consists of two cosine voltmeters and a phase splitter to give two signals 90° out of phase, and of equal magnitude. The phangle meter was developed to determine the impedance presented by an ear to an earphone.
24(1952); http://dx.doi.org/10.1121/1.1917423View Description Hide Description
For some time, engineers making noise measurements have required increasingly versatile equipment. The sound-level meter has been supplemented by accessories to meet this need, but it has frequently been a limiting factor in the over-all performance. To remedy this condition a new sound-level meter has been developed. It not only meets the standard specifications but also has the high fidelity range from 20 c to 20 kc necessary for utilization of modern microphones. The stabilized amplifier with wide dynamic range and separate output system supplies a signal suitable for frequency analyzers, high fidelity tape recorders, or level recorders. Miniaturization has led to a compact, light instrument. Many applications, however, can be handled by a simple noise meter and are economically feasible only if a low cost instrument is available. To meet this need a companion instrument, the sound-survey meter, has been developed, whose performance closely approximates that of standard sound-level meters but with simplified operating features. Its diminutive size and good characteristics make it ideal for many tests and survey measurements. With this low cost instrument available, more people will make acoustic measurements, which result should ultimately lead to more widespread understanding and better handling of acoustic problems.
24(1952); http://dx.doi.org/10.1121/1.1917424View Description Hide Description
There is now a great deal of evidence that the approximation of randomness in reverberation rooms used for absorptionmeasurements holds to a much greater extent than assumed in the classical analysis of the wave theory of sound in rooms by Morse and Bolt. Such factors as rotating vanes, averaging of measurements,diffusion due to wall absorption, and other elements, were not previously taken into account. Evidence of randomness includes (1) logarithmic decay curves linear over more than 60 decibels; (2) accurate random calibration of microphones at low frequencies; (3) measurement of correlation coefficients; and (4) fairly close fit between measured coefficients and those calculated by statistical theory from normal impedance measurements. The normal wall coefficient calculated by the perturbation theory given by Morse and Bolt gives coefficients as high as 6.00 for some materials while room measurements only slightly exceed statistical coefficients. If the random assumption holds, the chief source of error would appear to be the diffraction effects at the edges. The diffraction area correction for normal incidence has been derived from Morse and Bolt's treatment to be , where λ is the wavelength, σ the specific susceptance, L the length of free boundary, and S the area of the patch. It can be expected that the correction for random incidence would be greater. Using a percentage correction which is twice as great, good agreement is obtained between Riverbank coefficients and statistical coefficients over a wide range of frequencies and acoustic impedances.
24(1952); http://dx.doi.org/10.1121/1.1917425View Description Hide Description
It was the expressed opinion of the committees of the House and the Senate that “primary consideration should be given to the air conditioning, the lighting, and the acoustics of the remodelled Chambers.” This paper gives an account of the acoustical problems presented by the special requirements of rooms intended for deliberative assembly, together with the practical means used in the solution of these problems, and finally a summarized report of the acoustics of the two rooms as shown by tests made by Messrs. Wallace Waterfall and Hale J. Sabine.
24(1952); http://dx.doi.org/10.1121/1.1917426View Description Hide Description
Further study of decay characteristics in “dead” rooms with ultraspeed recording technique is made in several rooms using the method presented by the authors at the last meeting of the Society. The same studio is retested with the use of multiple tone sources consisting of semi-tone groups and chords. It is believed that the use of such sources saves the tedium of single tone explorations in locating peculiarities of decay.
24(1952); http://dx.doi.org/10.1121/1.1917427View Description Hide Description
Sounds in the low audiofrequency region produced by marine life have been recorded at the U. S. Navy SOFAR Station, Kaneohe Bay. The SOFAR hydrophones are located at a depth of 2100 feet on the eastern underwater slope on the island of Oahu. These sounds have levels of the order of 1 microbar, and many have a rather musical quality. There is a marked seasonal variation in the production of these sounds, with the early spring months being the period of most frequent occurrence. This variation is coincident with the seasonal variation of whales in the area, and this feature, plus the characteristics of the sounds themselves, has led to the belief that they are produced by whales—with possibly an assist or two from porpoises. The sounds are described, and information is presented on their spectrum, sound level, and frequency of occurrence.