Volume 48, Issue 1A, July 1970
Index of content:
- PROGRAM OF THE SEVENTY‐NINTH MEETING OF THE ACOUSTICAL SOCIETY OF AMERICA
- Session A. Psychological Acoustics I: Signal Detection and Loudness
- Contributed Papers
48(1970); http://dx.doi.org/10.1121/1.1912108View Description Hide Description
Plomp and Steeneken recently published results on the effect of phase on the timbre of complex tones consisting of 10 harmonically related frequency components [J. Acoust. Soc. Amer. 46, 409–421 (1969)]. The (rank order) similarity judgments obtained in triadic comparisons by eight subjects were represented geometrically in a multidimensional space using Kruskal's monotonic scaling procedure. In order to understand the observed effects of phase on timbre, it is assumed that the ear can detect in these comparison tests differences in the degree of amplitude modulation of all triplets of adjacent frequency components falling into the same critical band. Each 10‐tone complex is thus characterized by eight amplitude modulation indices (corresponding to the harmonic triplets 1,2,3; 2,3,4; 3,4,5; ⋯; 8,9,10). The Euclidean distances between two stimuli in this multidimensional “modulation space” are found to correlate well with the Euclidean distances in the perceptual space based on the similarity judgments.
48(1970); http://dx.doi.org/10.1121/1.1975093View Description Hide Description
Observers were asked to discriminate between a pair of 10‐μsec pulses and a single 20‐μsec pulse having the same total energy. The independent variable was the time, ΔT, between the two 10‐μsec pulses. The stimuli were also presented as elements in a periodic pulse train. The ΔT required for resolution of two clicks (two‐click threshold) was 10‐μsec. This value is in the same range as the separation needed to lateralize clicks delivered to the two ears and is about three orders smaller than the temporal separation needed to hear two clicks as distinctly separate. Whereas the addition of a steady background noise produced a remarkably small change in the magnitude of the two‐click threshold, performance deteriorated markedly when the pulses were lowpass filtered. It appears that discrimination of slight changes in the power spectrum of the two transient signals. especially in the high‐frequency region (8000 Hz and above) underlies the ear's sensitivity to a temporal discontinuity. [Research supported by the National Institute of Health, Public Health Service, U. S. Department of Health, Education, and Welfare, and by a research grant awarded to the Center for Human Information Processing.]
48(1970); http://dx.doi.org/10.1121/1.1975095View Description Hide Description
A critical duration was found, below which two acoustic bursts differing in their envelope curves are indiscriminable, and above which they sound different. The method of limits was used. Both bursts to be compared had the same energy, the same peak intensity, and the same carrier signal. The influence of the type of the carrier signal (tone or noise of different bandwidths), of its central frequency (250, 1000, and 4000 Hz), and of the shape of bursts on the critical duration was determined. Three shapes—Gaussian, divergent triangle (ramp up), and convergent triangle (ramp down)—were compared. With one exception the influence of factors mentioned above is small. The critical duration, expressed as the duration of a rectangular burst of the same energy and peak intensity, was found to be about 2 msec, except for the ramp up‐ramp down combination with the carrier at 4 kHz, for which the critical duration is 1 msec.
Discrimination of Sequential Constraints within Interval‐Coded Auditory Displays: Effect of Black Length48(1970); http://dx.doi.org/10.1121/1.1975097View Description Hide Description
Finite‐state sequences were translated to interval‐coded pulse trains, and converted to sound by earphones. In forced‐choice listening tests, listeners heard sequences with common elements, represented equally often over different block‐lengths. Interstate interval thresholds were obtained with different block‐lengths by an adaptive programming procedure. Such thresholds serve as an experimental measure of the discriminability between block‐lengths. Over an impressively wide range of conditions, interstate thresholds are closely related to the difference in informational uncertainty of the sequences.
48(1970); http://dx.doi.org/10.1121/1.1975099View Description Hide Description
At the previous meeting of this Society, we reported a lag effect in which the trailing syllable in a dichotic pair was more intelligible with time staggers at 15, 30, 60, and 90 msec. This paper presents data on 12 subjects who listened to time staggers of 90, 180, 250, and 500 msec. A special condition was also added in which the periodic portions of each pair were aligned without concern for the burst onset of the consonant. We called this the “boundary condition” and included it as a result of our observation that the voiceless consonant when paired with a voiced consonant is generally more intelligible in the dichotic condition. Both ears functioned about the same beyond 180‐msec staggers. However, we found the boundary condition enhanced the right ear laterality effect and attenuated the voiceless over voiced preponderance seen in ordinary dichotic simultaneous listening. [Supported by the NIH.]
48(1970); http://dx.doi.org/10.1121/1.1975101View Description Hide Description
Eight normal‐hearing male observers were trained to detect brief acoustic pips during a YES‐NO procedure until a consistent correct‐response rate of approximately 80% was achieved. The pips were presented to one ear at an equivalent level during reference, control, and experimental conditions. During the experimental conditions, the pips were presented at 90° re an on‐going 50‐Hz sinusoid, which was presented ipsilaterally at four subliminal levels (−10, −20, −30, and −40 db SL). Utilizing a balanced design, four observers received the signals in the right ear and four received them in the left. Higher correct‐response rates were observed when the 50‐Hz sinusoid was presented at −10 and −20 dB SL than during any of the other conditions. This increase in the correct‐response rate for the pips in the presence of a subliminal 50‐Hz sinusoid was called the Subliminal Sensitization Effect (SSE). When the effect of level was evaluated, it was noted that the SSE broke down when the 50‐Hz sinusoid was between −20 and −30 dB SL. Analyses of group differences revealed a greater SSE for the observers who received the signals in the right ear.
48(1970); http://dx.doi.org/10.1121/1.1975103View Description Hide Description
Likelihood ratio cutoff measures, β, were obtained on six mentally retarded adults (I.Q. 42–65) who participated in an extensive YES‐NOpsychoacoustical study. Experiments were designed within the framework of the theory of signal detectability and conducted under different values of optimum β by (1) systematically changing monetary values and costs associated with the decision outcomes and (2) changing prior probabilities of input occurrence. Of those observers yielding nonchance d′ measures, the observers who changed decision criteria in the direction of optimum β had higher I.Q. than those who did not. Of those observers who changed decision criteria, the change most often took place on the first day of the new condition. Measures of β on the second and third day were comparable to measures after several days. Indeed, later measures occasionally indicated that the observer was moving away from an optimum criterion.
48(1970); http://dx.doi.org/10.1121/1.1975105View Description Hide Description
Several models of the auditory system in which the signal flows along parallel channels were investigated. Some of the models studied exhibit nonlinear lateral interaction between channels. The effects of both single and multiple formant excitation and of intrinsic noise were considered. Performance was evaluated in terms of the signal‐to‐noise density ratio corresponding to a desired probability of correct detection given the presence of such signals for varying constraints on system information capacity, bandwidth, and complexity. Of particular significance to the analysis was the adoption of a fixed false‐alarm rate (Neyman‐Pearson relation) in favor of the common Bayes Criterion. A series of psychoacoustic tests were conducted, and the results of these tests, as well as published results by other investigators, were compared to the performance of the model. The data indicate that parallel‐channels models in general exhibit excellent noise performance and that, when used in conjunction with lateral interaction, they can successfully model many psychoacoustical phenomena.
48(1970); http://dx.doi.org/10.1121/1.1975107View Description Hide Description
The threshold of audibility of pure tones was measured by means of six psychophysical methods: adjustment, limits, constant stimuli, Békésy tracking, forced‐choice tracking, and confidence rating. The measurements were made at 250, 1000, and 4000 Hz for stimulus durations of 10, 25, 50, 100, 200, and 500 msec. The results obtained with the methods of adjustment, limits, constant stimuli, and confidence rating show a clear frequency dependence of the threshold shift as a function of stimulus duration—the threshold shift decreases as sound frequency increases. The results of the tracking methods do not show such an effect. An analysis of variance performed for the 10‐msec duration indicates a frequency effect at the 5% level of significance and an even more significant method‐frequency interaction. This outcome should resolve to some extent the issue of the frequency dependence of threshold‐duration characteristics.
48(1970); http://dx.doi.org/10.1121/1.1975109View Description Hide Description
The loudness summation of a pair of 12‐msec 1000‐Hz tone bursts was investigated as a function of the interburst time interval. The listeners were asked to match the loudness of a third burst to that of the burst pair or of the second burst. Both instructions yielded similar results. In an additional series of experiments, the first tone burst was replaced by a burst of broad‐band noise without materially changing the results. The loudness level plotted as a function of the time interval shows a minimum in the vicinity of 50 msec. At longer time intervals, it increases to a maximum at about 150 msec, then declines asymptotically to zero. The results may be accounted for by a combination of two processes: a linear summation with a time constant of 200 msec and an exponential poststimulatory recovery of neural sensitivity.
48(1970); http://dx.doi.org/10.1121/1.1975111View Description Hide Description
A binauralloudness balance was performed using subjects with normal hearing. The reference signal was a pulsed tone, and the balance signal was a steady tone of the identical frequency. The reference tone was presented at intensities of 20, 40, 60, and 80 dB re 0.0002 μbar. The balance tone consisted of a continuously varying intensity (increasing or decreasing) having an attenuation rate of 1, 2, 4, or 8 dB/sec, whose starting intensity was either suprathreshold or infrathreshold. As was anticipated, the different attenuation rates modified the balance point with the closest balance occurring with the use of 8 dB/sec attenuation rate. Suprathreshold starting intensities presented a poorer balance (balance at higher intensity) than did the infrathreshold starting intensities. An unexpected result was the differentiation of the balance levels of the supra‐ and infrathreshold starting intensities when compared with the various attenuation rates. A balance signal starting from below threshold gave results having a logarithmic relationship to attenuation rate, whereas a starting intensity above threshold yielded balance points having a linear relationship to that of the attenuation rate. It would appear that there are two distinct characteristics in the ear that are dependent upon the starting intensity to which the ear is exposed before performing a particular function.
48(1970); http://dx.doi.org/10.1121/1.1975113View Description Hide Description
In mathematics, the law of transitivity states quite simply that if A = B and if B = C, then A =C. The question of whether the law of transitivity holds for loudness level was investigated in this study to confirm an observation made in a previous report without adequate proof. [B. B. Bauer and E. L. Torick, “Researches in Loudness Measurement,” IEEE Trans. Audio Electroacoust. AU‐14, No. 3, 141–151. A panel of eight women were presented octave bands of pink noise from 125 Hz to 250 Hz to 2 kHz to 4 kHz, known to be at 70 phons loudness level, referred to a band of pink noise centered at 1 kHz. They were asked to compare the loudness of any octave band to each of the other bands also at 70 phons until all were compared. The stimuli were presented through a loudspeaker in a listening room with known acoustical properties. The results indicate that loudness level is indeed transitive. An analysis of variance is performed on the data to support the findings.
48(1970); http://dx.doi.org/10.1121/1.1975115View Description Hide Description
Direct electrical stimulation of the skin can arouse sensations similar to those from brief mechanical contact. In this study, tactile thresholds were obtained with brief pulses of direct current, under four different intensities of simultaneous broad‐band continuous noise ranging from 54 dB (estimated ambient noise) to 92 dB SL. Stimulation site was the right thenar eminence, the palmar base of the thumb. Electrical touch stimuli were 100‐msec‐long trains of 0.5‐msec anodal rectangular pulses, delivered by constant current stimulator at a pulse repetition rate of 100 pps. Thresholds were obtained from 12 subjects by means of a two‐alternative forced‐choice procedure. Stimulus intensity was varied by changing pulse amplitude alone. Electrotactile thresholds were highest under moderately intense noise, 66 and 79 dB SL. The magnitude of the threshold shift was small, less than 1.0 (tactile) dB, and statistically significant. Thresholds were lowest at both extremes of the sound intensity range, 54 and 92 dB SL. Variability around threshold was greatest under moderate intensity noise, least under ambient noise, where tactile thresholds were lowest. Electrotactile intensity discrimination, as it presumably involves stimulation behind the receptor, should measure limits for the touchsense. ΔI/I was most acute where absolute thresholds were worst, under moderate noise intensity. Conservative estimates of ΔI/I using run length are at worst 0.05/with ambient noise, a value in line with vibratory tactile measures.
- Session B. Architectural Acoustics I: Sound Transmission through Building Structures
48(1970); http://dx.doi.org/10.1121/1.1975117View Description Hide Description
The transmission of random incidence sound through single and double panels is studied using statistical energy analysis. In the single panel case the analytical model consists of three linearly coupled oscillators, room‐panel‐room. In the double panel case, the model consists of five oscillators, room‐panel‐cavity‐panel‐room. The effects of cavity absorption,panel damping,panel spacing, and edge conditions are included in the analysis. Both resonant and nonresonant transmission are considered. Theoretically, the sound energy transmitted is found to depend strongly on the radiation resistance of the panels and the panel and cavity loss factors. Good agreement is obtained between theory and experiment, and for double panels the agreement is much better than can be obtained using the London or Beranek and Work theories. The radiation resistance and vibration amplitude of the panels are also well predicted by the analysis. [Work conducted when both authors were at the Department of Building Science, The University, Liverpool, England.]
48(1970); http://dx.doi.org/10.1121/1.1975119View Description Hide Description
Cremer has shown that the damping factor of walls, the ratio of the imaginary to the real portion of their complex stiffness, is important in determining the sound transmission loss that a wall may afford above the frequency of coincidence of the trace of the acoustic wavelength upon the wall and the wavelength of the wall's flexural vibration. Measurements were made for several gypsum‐board and steel‐stud wall constructions. Many of these showed a constant value of the damping factor, but walls having fibrous insulation between studs or having a layer of wood fiberboard under the gypsum board showed an increased value of damping factor as the frequency was reduced below 4000 Hz. The measured value of damping factor was used to calculate the transmission loss and good agreement with measured TL was found.
48(1970); http://dx.doi.org/10.1121/1.1975121View Description Hide Description
This paper describes a procedure for calculating the impact‐noise level of the sound that would be radiated by a bare, stiff structural floor into the reverberant receiving room below, when excited with a standard tapping machine. Calculations are also given for the improvement in impact noise isolation achieved by adding: (a) an elasticsurface layer, (b) a locally reacting floating floor, and (c) a resonantly reacting floating floor, to the structural floor.
48(1970); http://dx.doi.org/10.1121/1.1975123View Description Hide Description
Based on the results presented in the preceding paper, an expression connecting the impact‐noise level (Ln ) and sound transmission loss (TL) of a floor is derived that takes into account both the resonant and the forced response of the floor; it thus extends the utility of the formulas derived previously by Heckl and Rathe into the frequency range below the critical frequency. It is shown how this connection between impact‐noise level and sound TL can be used for the experimental evaluation of the potential TL of a floating floor in the presence of certain types of flanking.
48(1970); http://dx.doi.org/10.1121/1.1975125View Description Hide Description
Five existing dormitory buildings at a large eastern U. S. university were surveyed in order to establish acoustical criteria for the university's campus housing building program. Objective acoustical measurements were made in each of the facilities, including background noise levels and noise reduction of walls, floors, and doors. In addition, subjective data were obtained distribution of questionnaires, which were returned by more than 600 students. Correlation of the subjective responses with the objective data is consistent with previously reported results on dormitories [J. A. Curtis and W. J. Cavanaugh, J. Acoust. Soc. Amer. 35, 787 (A) (1963)]. Sound isolation criteria are proposed for university level dormitories and construction guidelines are presented.
Comparison of the Aircraft Noise Reduction Achieved in Modified Houses in the United States and England48(1970); http://dx.doi.org/10.1121/1.1975127View Description Hide Description
Recently, five houses in three localities were modified in different ways to achieve more effective reduction of aircraft noise. (Results of modifications to two houses in Los Angeles were reported at the previous meeting.) The noise reductions achieved in the six rooms most extensively modified are compared with the noise reduction reported for 15 rooms in England. [W. E. Scholes and P. H. Parken, “The Insulation of Houses Against Noise from Aircraft in Flight,” J. Appl. Acoust.1, 37–46 (1968)]. The average noise reduction achieved in both the United States and English houses is quite similar. This leads to a tentative conclusion that the observed noise reductions for these modified rooms approach the maximum that can be achieved without very elaborate house modifications involving extensive changes in existing wall and roof constructions. [Studies supported by the Department of Housing and Urban Development.]
- Session C. Musical Acoustics I: Violins
- Invited Papers
48(1970); http://dx.doi.org/10.1121/1.1975129View Description Hide Description
While hologram interferometry is proving very useful in determining the plate resonances of string instruments such as guitars, viols, and violins, there are definite technical difficulties involved. Many of these relate to the extreme sensitivity of any interferometer to dimensional changes such as produced by wood expanding with humidity, or seismic noise. Others relate to mode combinations that often occur owing to close spacing in frequency of relatively low Q modes. This paper discussed hologram interferometry from a practical point of view, explaining those techniques which are most helpful in overcoming these difficulties. They include use of liquid gates and in‐place development of holograms to speed up real‐time measurements, 35‐mm film for recording time average holograms rapidly and in large numbers, the positioning of excitation points to separate combining modes, and laser speckle interferometry as an alternate to real‐time hologram interferometry. The measurement of Q values for vibration modes via hologram interferometry is also discussed.