利用残差网络分析普氏蹄蝠耳朵对恒频声源定向的影响

为研究恒频蝙蝠耳朵与空间定位的关系,利用深度学习算法和仿蝙蝠静态双耳接收器,分析蝙蝠耳朵对恒频声源定向的影响。首先根据普氏蹄蝠耳朵模型设计不同双耳夹角和间距的仿生双耳接收器,并从多个空间方位采集声源发射的不同频率的恒频声呐信号,然后提取双耳同步采集信号的时频图并归一化作为输入特征,最后利用残差网络实现声源定向。实验结果表明,静态双耳接收器对恒频声源的定向误差平均值基本保持在3.5°以下,但高于动态单耳接收器的定向误差;定向精度与声源频率及声源所在空间方位有关,声源位于接收器水平方向±30°范围内时,定向精度相对较高;双耳夹角和间距也会影响定向精度,且前者影响较为显著。     

A Comparison on the Localization Performance of Static and Dynamic Binaural Ambisonics Reproduction with Different Order

Ambisonics is a series of spatial sound reproduction system based on spatial harmonics decomposition and each order approximation of sound field. Ambisonics signals are originally intended for loudspeakers reproduction. By using head-related transfer functions (HRTFs) filters, binaural Ambisonics converts the Ambisonics signals for static or dynamic headphone reproduction. In present work, the performances of static and dynamic binaural Ambisonics reproduction are evaluated and compared. The mean binaural pressure errors across target source directions are first analyzed. Then a virtual source localization experiment is conducted, and the localization performances are evaluated by analyzing the percentages of front-back and up-down confusion, the mean angle error and discreteness in the localization results. The results indicate that binaural Ambsonics reproduction with insufficiently high order (for example, 5-10 order) is unable to recreate correct high-frequency magnitude spectra in binaural pressures, resulting in degradation in localization for static reproduction. Because dynamic localization cue is included, dynamic binaural Ambisoncis reproduction yields obviously better localization performance than static reproduction with the same order. Even a 3-order dynamic binaural Ambisoncis reproduction exhibits appropriate localizations performance.     

Summation bandwidths at threshold in normal and hearing-impaired listeners

The bandwidths for summation at threshold were measured for subjects with normal hearing and subjects with sensorineural hearing loss. Thresholds in quiet and in the presence of a masking noise were measured for complex stimuli consisting of 1 to 40 pure-tone components spaced 20 Hz apart. The single component condition consisted of a single pure tone at 1100 Hz; additional components were added below this frequency, in a replication of the G?ssler [Acustica 4, 408-414 (1954)] procedure. For the normal subjects, thresholds increased approximately 3 dB per doubling of bandwidth for signal bandwidths exceeding the critical bandwidth. This slope was less for the hearing-impaired subjects. Summation bandwidths, as estimated from two-line fits, were wider for the hearing-impaired than for the normal subjects. These findings provide evidence that hearing-impaired subjects integrate sound energy over a wider-than-normal frequency range for the detection of complex signals. A second experiment used stimuli similar to those of Spiegel [J. Acoust. Soc. Am. 66, 1356-1363 (1979)], and added components both above and below the frequency of the initial component. Using these stimuli, the slope of the threshold increase beyond the critical bandwidth was approximately 1.5 dB per doubling of bandwidth, thus replicating the Spiegel (1979) experiment. It is concluded that the differences between the G?ssler (1954) and Spiegel (1979) studies were due to the different frequency content of the stimuli used in each study. Based upon the present results, it would appear that the slope of threshold increase is dependent upon the direction of signal expansion, and the size of the critical bands into which the signal is expanded.     

Detection and discrimination of simulated motion of auditory targets in the horizontal plane

Three experiments investigated subjects' ability to detect and discriminate the simulated horizontal motion of auditory targets in an anechoic environment. "Moving" stimuli were produced by dynamic application of stereophonic balancing algorithms to a two-loudspeaker system with a 30 degree separation. All stimuli were 500-Hz tones. In experiment 1, subjects had to discriminate a left-to-right moving stimulus from a stationary stimulus pulsed for the same duration (300 or 600 ms). For both durations, minimum audible "movement" angles ("MAMA's") were on the order of 5 degrees for stimuli presented at 0 degrees azimuth (straight ahead), and increased to greater than 30 degrees for stimuli presented at +/- 90 degrees azimuth. Experiment 2 further investigated MAMA's at 0 degrees azimuth, employing two different procedures to track threshold: holding stimulus duration constant (at 100-600 ms) while varying velocity; or holding the velocity constant (at 22 degrees-360 degrees/s) while varying duration. Results from the two procedures agreed with each other and with the MAMA's determined by Perrott and Musicant for actually moving sound sources [J. Acoust. Soc. Am. 62, 1463-1466 (1977b)]: As stimulus duration decreased below 100-150 ms, the MAMA's increased sharply from 5 degrees-20 degrees or more, indicating that there is some minimum integration time required for subjects to perform optimally in an auditory spatial resolution task. Experiment 3 determined differential "velocity" thresholds employing simulated reference velocities of 0 degrees-150 degrees/s and stimulus durations of 150-600 ms. As with experiments 1 and 2, the data are more easily summarized by considering angular distance than velocity: For a given "extent of movement" of a reference target, about 4 degrees-10 degrees additional extent is required for threshold discrimination between two "moving" targets, more or less independently of stimulus duration or reference velocity. These data suggest that for the range of simulated velocities employed in these experiments, subjects respond to spatial changes--not velocity per se--when presented with a "motion" detection or discrimination task.     

Minimum audible angle thresholds for broadband noise as a function of the delay between the onset of the lead and lag signals

Minimum audible angle (MAA) thresholds were determined for six experienced subjects using a two-alternative, forced-choice adaptive paradigm. Broadband pink noise from a single generator was led to two identical speakers. The two sources were activated sequentially, each for a period of 10 ms. The subject's task was to indicate whether the second (lag) sound came from a source to the right or left of the first (lead) sound. The delay between the onset of the lead and the onset of the lag signal [interstimulus onset interval (ISOI)] was systematically varied from 1 ms (both 10-ms signals were concurrently active for 9 ms) to 200 ms. For a given ISOI, the spatial separation was varied adaptively to determine the MAA. A 450% improvement in auditory spatial resolution was evident as the ISOI increased from 1 to 150 ms. A further increase in the ISOI had no systematic effect on spatial resolution. These results indicate that there is a minimum integration period between 100-150 ms for the resolution of spatial information in the auditory modality.     

Minimum audible movement angle: marking the end points of the path traveled by a moving sound source

Four experienced subjects were tested on their ability to discriminate the direction of motion or the order of events in a single-interval, two-alternative, forced-choice adaptive paradigm. Two conditions, employing a broadband "pink" noise (500-8000 Hz), were examined: (1) A continuous noise was available from the moving sound source during the entire period of travel; and (2) 10-ms noise pulses were presented from the moving source at the beginning and end of the arc traveled (during the interpulse interval the source was inactive). Minimum audible movement angle (MAMA) thresholds were significantly lower when the moving source was active throughout the period of travel (0.914 degrees) than when only the end points of the arc of travel were "marked" (1.604 degrees). These results do not support the notion that the discrimination of motion can be reduced to a simple comparison of the location of the source at signal onset and the position of the source at signal offset. The MAMA thresholds obtained with broadband noise in the current experiment are considerably lower than the thresholds previously observed with tonal targets.     

Acoustic response and tuning in saccular nerve fibers of the goldfish (Carassius auratus)

The acoustic frequency selectivity of over 500 saccular nerve fibers of the goldfish was studied using automated threshold tracking based on spike rate increments defined statistically. Saccular fibers of the goldfish show great variation in (1) best sensitivity (-26 to + 35 dB re: 1 dyn/cm2), (2) best frequency (below 100 to 1770 Hz), (3) spontaneous rate (0 to over 200 spikes/s), (4) spontaneous type (silent, regular, irregular, burst), and (5) degree of tuning (Q 10 dB from less than 0.1 to 2). Saccular fibers may be grouped into four nonoverlapping categories based on tuning and best frequency: (1) untuned (less than 10-dB variation in sensitivity between 100 and 1000 Hz), (2) low frequency (BF from below 120 to 290 Hz), (3) midfrequency (BF between 330 and 670 Hz), and (4) high frequency (BF between 790 and 1770 Hz). Within each category, all spontaneous rates and types, and all degrees of tuning can be observed. The least sensitive fibers within each group have zero spontaneous rates. The goldfish is like all other vertebrates studied in that the peripheral auditory system is adapted for frequency selectivity throughout the animal's entire frequency range of hearing. Peripheral tuning most likely accounts for behavioral determinations of the "auditory filter" and for the detectability of signals masked by noise. The signal-to-noise ratio enhancement provided by these peripheral filters is likely to be of primary biological significance. A "place principle" of sound quality analysis based on lines "labeled" according to best frequency in the brain cannot be ruled out on the basis of the peripheral physiology.     

Precursor arrivals in the Yellow Sea, their distinction from first-order head waves, and their geoacoustic inversion

Measurements made as part of the 1996 Yellow Sea experiment at location 37 degrees N, 124 degrees E, undertaken by China and the U.S. are analyzed. Signals generated by explosive sources were received by a 60-m-length vertical line array deployed in waters 75 m deep. Evidence is presented that precursor arrivals measured at ranges less than 1 km are refracted waves that are zeroth order in their ray series classification, and this directly points to the existence of a gradient in sediment sound speed. In contrast, first-order head waves, which are much weaker in amplitude, would exist only if this gradient were absent. It is found that the energy spectrum of precursor arrivals agrees well with a zeroth-order model, i.e., it is proportional to the source amplitude spectrum, S(f), where f is frequency, rather than a first-order model, which would have it proportional to S(f)/f. From travel time analysis the sediment sound speed just below the water-sediment interface is estimated to be 1573 m/s with a gradient of 1.1 s(-1), and from analysis of the energy spectrum of the precursor arrivals the sediment attenuation is estimated to be 0.08 dB/m/kHz over the frequency range 150-420 Hz. The results apply to a nominal sediment depth of 100 m.     

Long-range sound propagation in the norwegian sea

The data of repeated experiments on the long-range propagation of explosion-generated and cw signals in the Norwegian Sea in summer conditions (with a fully-developed underwater sound channel) are presented. These data are used to analyze the spatial and time structures of the sound field, as well as to estimate the attenuation coefficient at frequencies within 63–630 Hz and to determine its frequency dependence. The spatial variability of the propagation conditions is analyzed on the basis of the experimental data obtained for the propagation of explosion-generated signals along a 815-km-long path crossing the Norwegian and Lofoten Hollows.     

Vibration and sound signatures of human footsteps in buildings

The acoustic signature of a footstep is one of several signatures that can be exploited for human recognition. Early research showed the maximum value for the force of multiple footsteps to be in the frequency band of 1-4 Hz. This paper reports on the broadband frequency-dependent vibrations and sound pressure responses of human footsteps in buildings. Past studies have shown that the low-frequency band (below 500 Hz) is well known in the literature, and generated by the force normal to the ground/floor. The seismic particle velocity response to footsteps was shown to be site specific and the characteristic frequency band was 20-90 Hz. In this paper, the high-frequency band (above 500 Hz) is investigated. The high-frequency band of the vibration and sound of a human footstep is shown to be generated by the tangential force to the floor and the floor reaction, or friction force. The vibration signals, as a function of floor coverings and walking style, were studied in a broadband frequency range. Different walking styles result in different vibration signatures in the low-frequency range. However, for the walking styles tested, the magnitudes in the high-frequency range are comparable and independent of walking style.     

Determination of the path of a sound source moving in an inhomogeneous medium

The possibility of determining the actual paths of moving sound sources from the signals received by a linear horizontal array whose size is large compared to the wavelength is investigated with the use of procedures suggested earlier for the imaging of dynamic objects moving below an inhomogeneous layer. Two cases of signal reception are considered: when the signals propagate in the oceanic waveguide and when the signals propagate through an inhomogeneous layer located near the array. It is shown that, unlike the standard spatial processing procedures, the proposed methods allow one to measure the absolute angular displacements to within the diffraction resolution of the array and to eliminate the ambiguity in angular measurements. An important point is that the proposed methods require no prior data on the parameters of the inhomogeneous layer or the multimode waveguide.     

Minimum audible movement angles as a function of sound source trajectory

Auditory resolution of moving sound sources was determined in a simulated motion paradigm for sources traveling along horizontal, vertical, or oblique orientations in the subjects's frontal plane. With motion restricted to the horizontal orientation, minimum audible movement angles (MAMA) ranged from about 1.7 degrees at the lowest velocity (1.8 degrees/s) to roughly 10 degrees at the highest velocity (320 degrees/s). With the sound moving along an oblique orientation (rotated 45 degrees relative to the horizontal) MAMAs generally matched those of the horizontal condition. When motion was restricted to the vertical, MAMAs were substantially larger at all velocities (often exceeding 8 degrees). Subsequent tests indicated that MAMAs are a U-shaped function of velocity, with optimum resolution obtained at about 2 degrees/s for the horizontal (and oblique) and 7-11 degrees/s for the vertical orientation. Additional tests conducted at a fixed velocity of 1.8 degrees/s along oblique orientations of 80 degrees and 87 degrees indicated that even a small deviation from the vertical had a significant impact on MAMAs. A displacement of 10 degrees from the vertical orientation (a slope of 80 degrees) was sufficient to reduce thresholds (obtained at a velocity of 1.8 degrees/s) from about 11 degrees to approximately 2 degrees (a fivefold increase in acuity). These results are in good agreement with our previous study of minimum audible angles long oblique planes [Perrott and Saberi, J. Acoust. Soc. Am. 87, 1728-1731 (1990)].(ABSTRACT TRUNCATED AT 250 WORDS)     

Spatial correlation of explosion-generated signals received by longitudinally separated hydrophones in the Mediterranean Sea

Results of an experiment carried out in the Mediterranean Sea on a propagation track ～400 km in length are presented. The sources of sound were 2.5-kg trinitrotoluene charges. The signals were received by three hydrophones spatially separated along the track by distances of 100 and 300 m. The hydrophones were carried by radio buoys, which were connected with each other and with the receiving vessel. The frequency band under study was 240–340 Hz. Under the conditions of the experiment, the spatial correlation interval obtained with averaging over the spreading time of multipath signals proved to be smaller than 100 m. It was shown that the resolution of signals arriving over individual ray paths or in narrow ray bundles considerably increased this interval. In this case, its experimental value was found to exceed 300 m.     

Correlation of underwater acoustic signals simultaneously received by omnidirectional and directional means

A comparison of experimental data on the spatial correlation between acoustic signals simultaneously received by an omnidirectional hydrophone and a directional vertical array is carried out. The spatial correlation was measured between the signals received at different distances in a deep ocean. The points of reception were positioned in two convergence zones along the path of sound propagation with a point-to-point distance of about 64 km. Pseudonoise signals were emitted in the frequency range (0.8–2.0) kHz and received by a vertical array, whose beam had a width of ∼2°. Concurrently, multipath signals received with the central hydrophone of the array were recorded. Signals in the first and second convergence zones were received at different times. Nevertheless, in the case of the directional reception, the coefficients of spatial correlation between such signals appeared to be as high as 0.64–0.74 even under the conditions of incomplete resolution of signals in the angle of arrival in the vertical plane. At the same time, in the case of omnidirectional reception, the coefficients of spatial correlation were below 0.32.     

Comprehensive studies of sound fields in the Kuril-Kamchatka region of the northwestern Pacific

The technique, experimental conditions, and main results of comprehensive studies of sound fields in the northwestern region of the Pacific Ocean are presented. The experiments are carried out on paths up to 2100 km in length. The power-frequency, space-time, and correlation characteristics of the sound fields are studied in sonic and infrasonic frequency bands for long-and extra-long-range propagation with the use of cw and explosion-generated sound signals. Effects of the bottom relief and the spatial distribution of the speed of sound on the frequency characteristics of the sound field are investigated. The role of front zones in the formation of sound fields received at the coastal shelf and in the open ocean is revealed. The loss coefficients are estimated. The space-time stability of the sound field components is studied, and the possibility is shown for the coherent components to be conserved and resolved in frequency at distances up to 2100 km. The phase velocities of these components are determined. The total broadening of the frequency spectra is considered. The correlation characteristics of the total field are obtained for horizontally separated receivers in sonic and infrasonic frequency bands.     

Minimum audible movement angle in the horizontal plane as a function of stimulus frequency and bandwidth, source azimuth, and velocity.

Minimum audible movement angles (MAMAs) were measured in the horizontal plane for four normal-hearing adult subjects in a darkened anechoic chamber. On each trial, a single stimulus was presented, and the subject had to say whether it came from a stationary loudspeaker or from a loudspeaker that was moving at a constant angular velocity around him. Thresholds were established by adaptively varying stimulus duration. In experiment 1, MAMAs were measured as a function of center frequency (500-5000 Hz), velocity (10 degrees-180 degrees/s), and direction of motion (left versus right). There was no effect of direction of motion. MAMAs increased with velocity, from an average of 8.8 degrees of arc for a target moving at 10 degrees/s to an average of 20.2 degrees of arc for a target moving at 180 degrees/s. MAMAs were higher for a 3000-Hz tone than for tones of lower or higher frequencies, as has been previously reported [D. R. Perrott and J. Tucker, J. Acoust. Soc. Am. 83, 1522-1527 (1988)]. In experiment 2, minimum audible angles (MAAs) were measured with sequentially presented stationary tone pulses (500-5000 Hz), and were shown to exhibit the same dependence on signal frequency that the MAMAs showed (average MAA at 3000 Hz: 8.4 degrees; average MAA at the other frequencies: 3.4 degrees). In experiment 3, MAMAs and MAAs were measured as a function of stimulus bandwidth (centered at 3000 Hz) and listening azimuth (0 degrees vs 60 degrees). Average MAAs decreased monotonically as stimulus bandwidth increased from 0 Hz to wideband (from 8.4 degrees to 1.2 degrees at 0 degrees azimuth; from 11.3 degrees to 1.5 degrees at 60 degrees azimuth). As in experiment 1, MAMAs increased with stimulus velocity, from values comparable to the MAAs for the slowest-velocity (10 degrees/s) targets to 70 degrees of arc or more in the poorest condition (third-octave band of noise presented at a velocity of 180 degrees/s and an azimuth of 60 degrees). MAMAs obtained in the slower-velocity conditions depended in the same way on stimulus bandwidth and listening azimuth that MAAs depended on these variables. In no case was the MAMA ever smaller than the MAA. It is hypothesized that a minimum integration time is required to achieve optimal performance in a dynamic spatial resolution task. Average estimates of this minimum time based on the current data vary from 336 ms (for targets presented at midline) to 1116 ms (for narrow-band targets presented at 60 degrees azimuth).(ABSTRACT TRUNCATED AT 400 WORDS)     

电力变压器绕组振动声纹特性分析*

为更加准确分析变压器绕组的状态特征,本文提出一种基于多物理场耦合仿真的变压器绕组振动声纹特性分析方法。根据实验条件,建立变压器绕组振动噪声模型,考虑变压器绝缘油在噪声传播过程中的作用,对S13-M-200/10型号的油浸式变压器进行短路实验,测量油箱表面的振动加速度以及周围空间的声音信号。仿真结果与实测数据对比分析,油箱表面的振动加速度集中频率为100Hz,空间声音信号集中频率为100Hz和200Hz,验证仿真模型的有效性。最后,建立变压器机械故障的仿真模型,分析得到变压器发生机械故障时,声音信号中100Hz频率分量减少,200Hz频率分量增加,为变压器绕组故障诊断提供依据。     

Evidence for different types of mechanoreceptors from measurements of the psychophysical threshold for vibrations under different stimulation conditions

The shape of the psychophysical frequency threshold curve for vibrations presented to the skin in the frequency region 5-1000 Hz is strongly dependent on the static force that the vibrator exerts on the skin and on whether there is a rigid surround around the vibrating contactor (presence of contrast). Where there is no rigid surround, an increase in static force reduces the threshold in the high-frequency region and increases it at low frequencies. When the static forces are sufficiently large, the thresholds reach a minimum value above 30 Hz and a maximum one below 30 Hz, this being the crossover frequency. Under these conditions in the frequency region around 200 Hz, where the threshold is determined by the Pacini receptor system, the vibration sensitivities of finger pad and thenar eminence (glabrous skin) are equal, while the value for the inner side of the forearm (hairy skin) is 12 dB higher. However, when a rigid surround is used, the threshold increases above 30 Hz and decreases below 30 Hz. The latter increase in sensitivity, which is introduced by the presence of contrast cues, amounts to about 20 dB and is sharply tuned at 18 Hz for the glabrous skin of the finger. It is argued that in this case the threshold is determined by the Meissner receptor system. This increase in sensitivity is less pronounced (about 10 dB) and less sharply tuned for the other sites. Finally, when the contact of the vibrating surface to the skin is at a minimum, the vibration threshold has the same displacement value (about 3 microns) over the whole frequency region independent of the site of stimulation and whether or not a rigid surround is present.     

Influence of the Number of Loudspeakers on the Timbre in Horizontal and Mixed-Order Ambisoncis Reproduction

Ambisonics is a series of flexible spatial sound reproduction systems based on spatial harmonics decomposition of sound field. Traditional horizontal and spatial Ambisonics reconstruct horizontal and spatial sound field with certain order of spatial harmonics, respectively. Both the Shannon-Nyquist spatial sampling frequency limit for accurately reconstructing sound field and the complexity of system increase with the increasing order of Ambisonics. Based on the fact that the horizontal localization resolution of human hearing is higher than vertical resolution, mixed-order Ambisonics (MOA) reconstructs horizontal sound field with higher order spatial harmonics, while reconstructs vertical sound field with lower order spatial harmonics, and thereby reaches a compromise between the perceptual performance and the complexity of system. For a given order horizontal Ambisoncis or MOA reproduction, the number of horizontal loudspeakers is flexible, providing that it exceeds some low limit. By using Moore’s revised loudness model, the present work analyzes the influence of the number of horizontal loudspeakers on timbre both in horizontal Ambisonics and MOA reproduction. The binaural loudness level spectra (BLLS) of Ambisoncis reproduction are calculated and then compared with those of target sound field. The results indicate that below the Shannon-Nyquist limit of spatial sampling, increasing the number of horizontal loudspeakers influence little on BLLS then timbre. Above the limit, however, the BLLS for Ambisoncis reproduction deviate from those of target sound field. The extent of deviation depends on both the direction of target sound field and the number of loudspeakers. Increasing the number of horizontal loudspeakers may increase the change of BLLS then timbre in some cases, but reduce the change in some other cases. For MOA, the influence of the number of horizontal loudspeakers on BLLS and timbre reduces when virtual source departs from horizontal plane to the high or low elevation. The subjective evaluation experiment also validates the analysis.     

Where the ocean influences the impulse response and its effect on synchronous changes of acoustic travel time

In 1983, sounds at 133 Hz, 0.06 s resolution were transmitted in the Pacific for five days at 2 min intervals over 3709 km between bottom-mounted instruments maintained with atomic clocks. In 1989, a technique was developed to measure changes in acoustic travel time with an accuracy of 135 microseconds at 2 min intervals for selected windows of travel time within the impulse response. The data have short-lived 1 to 10 ms oscillations of travel time with periods less than a few days. Excluding tidal effects, different windows exhibited significant synchronized changes in travel time for periods shorter than 10 h. In the 1980s, this phenomenon was not understood because internal waves have correlation lengths of a few kilometers which are smaller than the way sound was thought to sample the ocean along well-separated and distinct rays corresponding to different windows. The paradox's resolution comes from modern theories that replace the ray-picture with finite wavelength representations that predict sound can be influenced in the upper ocean over horizontal scales such as 20 km or more. Thus, different windows are influenced by the same short-scale fluctuations of sound speed. This conclusion is supported by the data and numerical simulations of the impulse response.