Volume 90, Issue 5, November 1991
Index of content:
90(1991); http://dx.doi.org/10.1121/1.402043View Description Hide Description
This article describes an experimental study of the motion of the grand piano key and hammer. The motions were registered by means of an optical position measuringsystem. The measurements include typical key and hammer motions at different dynamic levels and for different types of ‘‘touch.’’ Also, the accelerating force on the hammer before release (‘‘let‐off’’), and the following free motion of the hammer were investigated. Resonances in the hammer were detected and examined by modal analysis. A possible influence of the pianist’s touch on the string vibrations via the hammer resonances is discussed.
90(1991); http://dx.doi.org/10.1121/1.402044View Description Hide Description
The purpose of this study was to examine several factors of vocal quality that might be affected by changes in vocal fold vibratory patterns. Four voice types were examined: modal, vocal fry, falsetto, and breathy. Three categories of analysis techniques were developed to extract source‐related features from speech and electroglottographic (EGG) signals. Four factors were found to be important for characterizing the glottal excitations for the four voice types: the glottal pulse width, the glottal pulse skewness, the abruptness of glottal closure, and the turbulent noise component. The significance of these factors for voice synthesis was studied and a new voice source model that accounted for certain physiological aspects of vocal fold motion was developed and tested using speech synthesis. Perceptual listening tests were conducted to evaluate the auditory effects of the source model parameters upon synthesized speech. The effects of the spectral slope of the source excitation, the shape of the glottal excitation pulse, and the characteristics of the turbulent noise source were considered. Applications for these research results include synthesis of natural sounding speech, synthesis and modeling of vocal disorders, and the development of speaker independent (or adaptive) speech recognition systems.
90(1991); http://dx.doi.org/10.1121/1.402045View Description Hide Description
Hearing threshold levels have been determined in the low‐frequency range (20–500 Hz) on a group of 30 young normal‐hearing subjects using monaural stimulus presentation through an insert earphone (Etymotic Research ER‐3A). A retest was performed on half of the group to provide data on test–retest reliability. The mean hearing threshold levels obtained agree closely with the Minimum Audible Field data of ISO 226, however, with some deviation at the very lowest frequencies below 40 Hz. The test–retest difference results yielded mean values that averaged 1.15 dB with an average standard deviation across test frequencies of 3.9 dB. The results show that low‐frequency hearing thresholds for pure tones of frequencies from 40 Hz and up can be determined with acceptable validity and reliability by the use of this type of insert earphone.
90(1991); http://dx.doi.org/10.1121/1.402046View Description Hide Description
Observations of the otoacoustic emissions(OAEs) evoked during a continuous single stimulus tone have been made on humans using a nonlinear residual time domain technique. The technique, described in this paper, involved the digital summation of responses to contiguous stimulation intervals, some of which included short bursts of a suppressor, or probe, tone. Stimulus intervals are constructed so that both the stimulus and probe tones summed to zero cyclically, leaving a residual response. This residual is attributable to the nonlinearity of the whole acoustic response, as measured in the ear canal, to the stimulus and probe tone complex. A theoretical treatment of this paradigm is presented examining the relation of this residual to the OAE evoked by the stimulus tone. It is shown experimentally that the residual, found at the stimulus tone frequency, has a latency and a saturating input–output growth functions indicative of an OAE. The detailed OAE amplitude‐versus‐frequency variations, and the general latencies of the OAEs in two human ears were measured using both the constant tone evoked residual method described and the click evoked delayed emission method. The results from both methods are in agreement. The frequency‐dependent properties of the suppression of the OAE were investigated using various stimuli to probe frequency ratios. The continuous tone time domain residual method has advantages for the observation of stimulus frequency OAEs and for relating these to any distortion product simultaneously generated.
Spike discharge properties that are related to the characteristic frequency of single units in the frog auditory nerve90(1991); http://dx.doi.org/10.1121/1.402047View Description Hide Description
Single units from the auditory nerve of frogs and toads have their receptor cells located in two separate sensory organs that provide disjoint frequency ranges. The amphibian papilla (ap) provides units with characteristic frequency (CF) in the low‐ and mid‐frequency regions and the basilar papilla (bp) provides units with high CF. There are gross differences in both the mechanical design and innervation patterns of the two organs, so that one might expect discharge properties for units with different CF to differ in many respects. However, there have been few reports of response attributes that correlate strongly with CF for units in the mid‐ and high‐CF regions. Measurements of automated tuning curves from 250 units in R a n a p i p i e n s show that W 10 d B , the bandwidth of the tuning curve measured 10 dB above CF threshold, is consistently larger for high‐CF units than for low‐ and mid‐frequency units. When units are classified into three groups by an objective statistical method using only CF and W 10 d B measurements, the groups appear to correspond reasonably well with the low‐, mid‐, and high‐frequency categories identified in many other studies.
90(1991); http://dx.doi.org/10.1121/1.402048View Description Hide Description
Human and chinchilla listeners exhibit nonmonotonic temporal acuity for speechsounds differing in voice onset time (VOT). Characteristics of the neural discharge pattern or of the stimuli themselves that might account for the pattern of temporal acuity have not been described. Responses of chinchilla auditory‐nerve fibers to syllables from an alveolar VOT continuum were measured. Peak discharge rates and peak response latencies elicited by the syllables with the shortest and longest VOTs were highly variable across groups of neurons with similar characteristic frequencies. For VOTs from the middle of the continuum, peak responses were larger, and response latencies were nearly constant across the same group of neurons. Overall, the magnitude and temporal variability of the responses of populations of primary auditory neurons varied nonmonotonically with VOT, consistent with the pattern of psychophysical temporal acuity for these syllables exhibited by humans and chinchillas. Spectral analyses suggested by the pattern of neural responses indicated that synchronous or correlated spectral cues were available over a wider bandwidth for those syllables from the middle of the continuum for which the neural representation was least variable.
90(1991); http://dx.doi.org/10.1121/1.402049View Description Hide Description
This paper describes the temporal responses of anteroventral cochlear nucleus (AVCN) units in the chinchilla to rippled noises. Rippled noise is generated when a wideband noise is delayed and added (cos+ noise) or subtracted (cos− noise) to the undelayed noise. Renewal densities were constructed to evaluate synchronous discharges at the delay. In response to rippled noise, AVCN units which show phase locking to best frequency (BF) tones gave renewal densities having major peaks at the delay for cos+ noise, but nulls at the delay for cos− noise. Most AVCN units which did not show BF phase locking gave renewal densities that did not contain features related to the rippled noise delay; a few of these nonphase‐locked units did show peaks in renewal densities for both cos+ and cos− noises. Synchrony at the rippled noise delay was also demonstrated with evoked potential recording. Autocorrelation functions of the neurophonic potential showed peaks at the rippled noise delay for both cos+ and cos− noises. In addition, peaks could be observed in the autocorrelation functions of neurophonic potentials for rippled noises with delays as short as 1 ms; peaks were never observed in renewal densities of single units for ripple delays as short as 1 ms. The results show that a temporal representation of rippled noise delay does exist in the AVCN and are consistent with current hypotheses regarding functions of AVCN subsystems. The temporal representation of the delay is a presumptive neural code for the pitches of rippled noises.
90(1991); http://dx.doi.org/10.1121/1.402050View Description Hide Description
Steady‐state evoked potential responses were measured to binaural amplitude‐modulated (AM) and combined amplitude‐ and frequency‐modulated (AM/FM) tones. For awake subjects, AM/FM tones produced larger amplitude responses than did AM tones. Awake and sleeping responses to 30‐dB HL AM/FM tones were compared. Response amplitudes were lower during sleep and the extent to which they differed from awake amplitudes was dependent on both carrier and modulation frequencies. Background EEG noise at the stimulus modulation frequency was also reduced during sleep and varied with modulation frequency. A detection efficiency function was used to indicate the modulation frequencies likely to be most suitable for electrical estimation of behavioral threshold. In awake subjects, for all carrier frequencies tested, detection efficiency was highest at a modulation frequency of 45 Hz. In sleeping subjects, the modulation frequency regions of highest efficiency varied with carrier frequency. For carrier frequencies of 250 Hz, 500 Hz, and 1 kHz, the highest efficiencies were found in two modulation frequency regions centered on 45 and 90 Hz. For 2 and 4 kHz, the highest efficiencies were at modulation frequencies above 70 Hz. Sleep stage affected both response amplitude and background EEG noise in a manner that depended on modulation frequency. The results of this study suggest that, for sleeping subjects, modulation frequencies above 70 Hz may be best when using steady‐state potentials for hearing threshold estimation.
The focus measurement technique for estimation of arbitrary time delays in multichannel, multievent systems90(1991); http://dx.doi.org/10.1121/1.402051View Description Hide Description
The estimation of arbitrary time delays in multichannel, multievent systems is studied using a new method called the focus measurement technique (FMT). The method introduced uses the window concept to isolate segments of the data record. However, instead of calculating the generalized cross correlation to estimate delays between data segments, the FMT (based on the assumption of normal distribution) combines windowed‐segment points of every two channels to obtain the focal length of an ellipse which contains a certain amount of these distributed points. Based on simple geometric analysis, the FMT utilizes a windowed‐segment distribution in order to estimate the event delays in the system. Simulation results show that this technique can find delays very accurately down to a signal‐to‐noise ratio (SNR=mean‐square value of signal/mean‐square value of noise) of 0 dB while reducing the amount of computation involved.
90(1991); http://dx.doi.org/10.1121/1.402052View Description Hide Description
A method of source localization in shallow water, based on subspace concept, is described. It is shown that a vector representing the source in the image space spanned by the direction vectors of the source images is orthogonal to the noise eigenspace of the covariance matrix. Computer simulation has shown that a horizontal array of eight sensors can accurately localize one or more uncorrelated sources in shallow water dominated by multipath propagation.
90(1991); http://dx.doi.org/10.1121/1.402053View Description Hide Description
In this paper, the problem of signal detection in severe and/or changing noise environments, often encountered in underwater acoustics and communications, is considered. The detectors operate in sequential, i.e., variable sample size, mode at near optimum levels for a particular noise environment and are robust by maintaining high efficiency in other than the nominal noise environments by adapting their optimum nonlinearity using an m‐interval polynomial approximation (MIPA) of it. Furthermore, the assumption of independent samples is relaxed, allowing higher transmission rates for lower error rates. The proposed sequential detectors work well for coherent and noncoherent detection, are asymptotically optimum and, for small signal‐to‐noise ratios (SNR), increase their transmission rate up to four times as compared to their fixed‐sample size counterparts. The performance of the detectors is evaluated in typical underwater noise fields. It is demonstrated that their efficiency with respect to sequential MIPA detectors with independent sampling is improved, in some cases up to the order of the dependence of the samples Q. It is emphasized that the MIPA detectors are easy to implement. The estimation and updating of the detector parameters may be accomplished using parallel processors operating in a recursive mode without disturbing the decision process.
90(1991); http://dx.doi.org/10.1121/1.402054View Description Hide Description
Active minimization of total power output and active absorption of sound power are analyzed, using a general impedance‐based approach, for an array of controllable secondary sources and an array of original primary sources. When the total power output of the two arrays is minimized, and the primary source array is all in phase, the power output of each of the secondary sources is found to be exactly zero. When the power absorption of the secondary source array is maximized, the net power output of the primary source array can be either reduced or increased, compared to that in the absence of control, depending on the properties of the transfer impedances. If the primary and secondary sources are well coupled (as is the case when they are spaced less than a quarter wavelength apart in free‐space or when they are in an enclosure excited near the natural frequency of a lightly damped mode) minimizing the total power output gives worthwhile reductions in the radiated power of the primary source. Maximizing the power absorption of the secondary source under these conditions, however, can considerably increase the power output of the primary source. If the sources are well separated, compared with the wavelength, in the free‐field, or are placed in a diffuse sound field in an enclosure, the coupling between the sources is weak. Under these conditions the secondary source can effect little reduction in the power output of the primary source. Although acoustic power can still be absorbed by the secondary source under such circumstances, the amount of absorbed power will always be small compared with the power output of the primary source.
90(1991); http://dx.doi.org/10.1121/1.402055View Description Hide Description
Bies and Hansen [J. Acoust. Soc. Am. 8 8, 2743–2754 (1990)] have proposed an alternative formulation of the relationship between noise exposure and noise‐induced hearing impairment to that presented in International Standard ISO 1999, in which they assume that presbycusis and noise‐induced permanent threshold shift (NIPTS) are additive on an antilogarithm basis. Data concerning deterioration in hearing threshold levels at 4000 Hz due to aging in war veterans with NIPTS do not support the Bies and Hansen assumption but provide support for the formula for combining presbycusis and NIPTS incorporated in International Standard ISO 1999.
90(1991); http://dx.doi.org/10.1121/1.402056View Description Hide Description
Criteria based on Zwicker’s model of loudness and the A‐weighted sound‐pressure level are used to find the source–receiver distance at which atmospheric absorption becomes significant. The problem is studied for a wide range of temperature and relative humidity variations. It is found that both criteria are nearly equivalent. For the majority of environmental noise sources, including motor vehicles and trains, air absorption cannot be neglected at distances beyond a few hundred meters.
90(1991); http://dx.doi.org/10.1121/1.402057View Description Hide Description
A technique for deriving the optimal surface velocity distribution on the surface of a finite baffled beam has been developed. The optimal velocity distribution minimizes the radiation efficiency of the beam for a specified maximum permissible mode and frequency. A modal expansion of the surface velocity in terms of unknown modal amplitude coefficients, the Rayleigh integral, and a far‐field intensity integration are employed to obtain a quadratic expression for the radiation efficiency of the beam. Application of a suitable constraint to avoid trivial solutions leads to an eigenvalue problem identical in form to the Rayleigh quotient employed in dynamic mechanical systems. The eigenvector of modal amplitude coefficients corresponding to the lowest eigenvalue yields the minimum radiation efficiency, while the eigenvalue itself is the actual value of the minimum radiation efficiency. Near and below coincidence, the optimal eigenvector of modal amplitude coefficients yields a radiation efficiency significantly less than the radiation efficiency of any single modal component acting alone. Simply supported and clamped–clamped boundary conditions are considered, and numerical examples are presented for each.
Temperature‐rate‐dependent thermoelastic waves in a homogeneous plate due to a suddenly punched hole90(1991); http://dx.doi.org/10.1121/1.402058View Description Hide Description
The distribution of deformation, temperature, and stress in a homogeneous isotropic, thermally conducting, infinitely extended, stretched elastic plate due to a flat nose cylindrical projectile has been studied in the context of the Green–Lindsay theory of thermoelasticity. As the ‘‘second sound’’ effects are short lived, the discussion is confined to small‐time approximations. The short‐time solutions have been obtained by using the Laplace transforms technique. The deformation is found to be continuous but the temperature and stress are found to be discontinuous. The jumps obtained have been computed numerically and are represented graphically for carbon‐steel material.
90(1991); http://dx.doi.org/10.1121/1.402059View Description Hide Description
The scattering of sound waves by an air‐filled, elastic, spherical shell in deep waters, in the frequency domain is analyzed. This exact analysis is based on the classical formulation of three‐dimensional elastodynamics, for fluid‐loaded spherical shells of arbitrary thickness. Form functions, residual responses, and the partial‐wave expansions of both, are determined. Results are displayed in relatively wide frequency bands for various increasing shell thicknesses. The resonance features in the sonar cross sections (SCS) are isolated by means of a new hybrid modal background that substantially improves the results found with the earlier (rigid/soft) backgrounds of the resonance scattering theory (RST). The resonance features in the SCSs that correspond to each mode, and also to each of the various shell waves that propagate around its periphery are isolated. There seem to be over half‐a‐dozen shell (generalized Lamb and Stoneley) waves manifesting their influence in the SCSs within the examined band. Three large‐amplitude features are most noticeable. There is a (slow) wave due to the double curvature of the shell which is responsible for a large, spiky feature at low frequencies.
There is the spherical A 0 wave, caused by the coincidence effect, responsible for a broad ‘‘bump’’ with superimposed spikes, which appears in the midfrequency region, near x∼(h’)−1, where h’ is the relative shell thickness. There is also a thickness resonance feature at high frequencies, which is due to a wave that propagates through the shell thickness at the incidence point. There is a spherical Stoneley‐type wave that resides mostly in the water in the frequency region below coincidence, x c , at which the A 0 wave is activated. Finally, there are two resonance families caused by spherical analogs of the symmetric, S n , and the antisymmetric, A n , Lamb waves originally studied in flat plates. These spherical generalizations are denoted by the same symbols S n , A n , with the understanding that they now refer to shells. This model’s results for phase velocities, cutoff frequencies, etc..., are compared to those produced by simpler approaches in order to size the approximations introduced by those simpler models. This serves to establish the benchmark nature of the present model and of the calculations it produces, which are displayed in many instances. Dispersion plots are generated for the phase velocities c p (x) of all the above‐mentioned shell waves. These c p (x) are proportional to the (real parts of the) roots of determinantal conditions that are also analyzed. Critical angles, coincidence phenomena, and reflection and transmission properties of the shell are examined and all the required physical interpretations are given.
90(1991); http://dx.doi.org/10.1121/1.402348View Description Hide Description
In this paper an alternative method to first threshold crossing for acoustic emission(AE)source location is presented. For wave propagation in dispersive media, the accuracy of source location can be improved by locating corresponding phase points on the transducer outputs to determine the difference in arrival times. The phase point location was done by cross‐correlating the transducer outputs with a single frequency cosine wave modulated by a Gaussian pulse. Experiments were performed using a lead break as the AEsource on the surface of an aluminum plate. Due to the plate geometry and source orientation, the wave produced was highly dispersive. Although this wave was unsuitable for first threshold crossing techniques, the time differences needed for triangulation could be determined using the cross‐correlation technique.
Analysis of elastic wave propagation in stratified fluid‐filled porous media for interwell seismic applications90(1991); http://dx.doi.org/10.1121/1.402060View Description Hide Description
The analytical solution for seismic wave propagation associated with a point force in a fluid‐filled porous medium is developed. The point force solution is applied to solve the boundary value problem of seismic wave propagation in a stratified poroelastic medium. The coupled Biot vector wave equations are expressed in cylindrical coordinates and expanded in Fourier series with respect to azimuth. The resulting equations are transformed to the wave‐number domain using the Hankel transform method. Following this analysis, one set of three coupled partial differential equations associated with fast compressional, slow compressional, and vertically polarized shear waves is derived. The unknowns are the radial and vertical displacements associated with the solid frame motion as well as the fluid pore pressure. A separate partial differential equation associated with waves whose particle motion is polarized in horizontal planes (S H waves) is derived as well. The general solution of the three coupled differential equations is obtained by the Kupradze method.
This solution of the Biot’s motion equations in the wave‐number domain leads directly to closed form expressions for the vector wave displacement and the pressure produced by a point force in a poroelastic unbounded medium. In order to develop the solution of a point force in the presence of a stratified porous medium, the displacement‐stress matrix, the pressure, and the vertical component of the displacement of the fluid in its relative motion versus the solid for different regions are expressed in terms of upgoing and downgoing waves and unknown wave coefficients. The wave coefficients are determined by applying boundary conditions of continuity of displacements, pressure and stresses across each layer interface, and the radiation conditions at infinity. To determine the unknown wave coefficients, a method that consists in expressing the kernels of the Hankel transform integrals in terms of factorization of upgoing and downgoing wave amplitudes in each layer is used. These factorizations are based on the generalized reflection and transmission coefficient matrices, which are formed recursively, from one layer boundary to the next, including all of the reflection/conversion/transmission properties of the layered medium.
Their factorization method allows the field within each layer above or below the source to be determined once the field in the medium containing the source is known. The final equations provide a complete description of the field throughout the layered medium. The particular form of the equations makes possible the simultaneous evaluation of the response at a number of detector locations for a number of different source positions in a borehole for interwell seismic applications. Numerical model results demonstrate the validity of this theoretical development for predicting spectral responses associated with porosity and permeability effects. The seismic pressure response of a thin gas‐saturated porous layer was analyzed. The results inferred that the gas‐saturated porous layer strongly attenuates the waveforms observed by detectors within the layer. Alternatively, large multiple reflections and converted waves from the layer are observed by detectors in the water‐saturated porous formation.
Effects of random deviations in interface properties on the propagation of ultrasound in thick composites90(1991); http://dx.doi.org/10.1121/1.402061View Description Hide Description
Propagation of ultrasound across a solid layer with equally spaced parallel interfaces is studied by using a transfer matrix method. For a layer with identical interfaces the propagation of ultrasound is governed by a dispersion relation, which displays passing and stopping bands in the frequency domain. For a layer with interfaces that have random deviations from mean interface properties, the Fürstenberg theorem and Monte Carlo simulation have been used to study propagation of ultrasound across the layer. It has been shown that at all frequencies there is an exponential decay in the amplitude of the transmitted wave. This decay, which increases with increasing frequency, defines a localization phenomenon, since wave motion will be confined to the insonified side of the layer.