Linking the sounds of dolphins to their locations and behavior using video and multichannel acoustic recordings

It is difficult to attribute underwater animal sounds to the individuals producing them. This paper presents a system developed to solve this problem for dolphins by linking acoustic locations of the sounds of captive bottlenose dolphins with an overhead video image. A time-delay beamforming algorithm localized dolphin sounds obtained from an array of hydrophones dispersed around a lagoon. The localized positions of vocalizing dolphins were projected onto video images. The performance of the system was measured for artificial calibration signals as well as for dolphin sounds. The performance of the system for calibration signals was analyzed in terms of acoustic localization error, video projection error, and combined acoustic localization and video error. The 95% confidence bounds for these were 1.5, 2.1, and 2.1 m, respectively. Performance of the system was analyzed for three types of dolphin sounds: echolocation clicks, whistles, and burst-pulsed sounds. The mean errors for these were 0.8, 1.3, and 1.3 m, respectively. The 95% confidence bound for all vocalizations was 2.8 m, roughly the length of an adult bottlenose dolphin. This system represents a significant advance for studying the function of vocalizations of marine animals in relation to their context, as the sounds can be identified to the vocalizing dolphin and linked to its concurrent behavior.     

Analysis and localization of blue whale vocalizations in the Solomon Sea using waveform amplitude data

During the Woodlark Basin seismic experiment in eastern Papua New Guinea (1999-2000), an ocean-bottom seismic array recorded marine mammal vocalizations along with target earthquake signals. The array consisted of 14 instruments, 7 of which were three-component seismometers with a fourth component hydrophone. They were deployed at 2.0-3.2 km water depth and operated from September 1999 through February 2000. While whale vocalizations were recorded throughout the deployment, this study focuses on 3 h from December 21, 1999 during which the signals are particularly clear. The recordings show a blue whale song composed of a three-unit phrase. That song does not match vocalization characteristics of other known Pacific subpopulations and may represent a previously undocumented blue whale song. Animal tracking and source level estimates are obtained with a Bayesian inversion method that generates probabilistic source locations. The Bayesian method is augmented to include travel time estimates from seismometers and hydrophones and acoustic signal amplitude. Tracking results show the whale traveled northeasterly over the course of 3 h, covering approximately 27 km. The path followed the edge of the Woodlark Basin along a shelf that separates the shallow waters of the Trobriand platform from the deep waters of the basin.     

全光光纤水听器系统海上试验

阐述了光纤水听器拾取声压信号的基本原理。研制的光纤水听器探头尺寸为Φ22 mm×100 mm,相位灵敏度为-160 dB(参考值1 rad/μPa),波动小于3 dB,系统最小可测声压在1 kHz频点为43.5 dB,与零级海洋噪声水平相当。光纤水听器与标准水听器采集的宽带信号相关系数大于0.94,信号基本一致。23基元1 km传输全光湿端光纤水听器阵列于2002年8月在渤海进行了海上试验,相邻基元采集的气枪信号相关系数大于0.98,阵列具有良好波束指向性。光纤水听器关键技术的成熟为其走向应用奠定了良好基础。     

深海脉冲信号簇到达结构特征及其在水下声源定位中的应用

为了实现对深海水下声源的定位,对典型深海环境中的脉冲信号到达结构特征进行了理论分析,给出了声源和接收位置位于近海面时到达信号的簇信号形式近似表达式。当声源和接收位置处于近海面深度时,接收到的信号呈簇状结构形式。在提取到达信号中各簇信号到达时间、幅值等特征参数的基础上,提出了一种通过对到达信号中簇信号特征参数匹配搜索进行水下脉冲信号定位的方法,仿真分析了不同距离和不同深度处的簇信号特征,并利用一次南海海域爆炸声源的声传播实验数据进行了实验验证。结果表明:各簇信号对应的到达时间、幅值等特征参数可用于对声源位置的匹配定位;海上实验中利用单水听器得到的水下声源的距离估计结果与实测距离结果较为一致,对实验中2.0 km至90.4 km内声源的距离估计误差不大于8%。     

Experimental analysis on air-to-water sound transmission loss in shallow water

The research of propagation characteristics of air-to-water sound transmission is of great importance to the detection of aerial targets from underwater.In order to study the propagation characteristics of air-to-water sound transmission in shallow water,State Key Laboratory of Acoustics,Institute of Acoustics,conducted an experiment in the South China Sea in March,2013.During the experiment,multi-frequency signals transmitted by a hooter hung on a research ship were received by an underwater hydrophone,and the distance between the hooter and the hydrophone was from 2.4 km to 9.8 km approximately.Through analyzing experimental data in this work,the experimental air-to-water transmission loss at frequencies128 Hz and 256 Hz is estimated up to 9.8 km in range,and its oscillation structure is evident.The wave-number integration approach is used to simulate theoretical air-to-water transmission losses,which are in good agreement with experimental values and to explain the experimental air-to-water sound transmission characteristics.     

空气中声源激发浅海水下声场传播特性的实验研究

为了获得空气中远距离声源激发水下声场的精细结构,2013年3月,声场声信息国家重点实验室在南海海域进行了一次空气中声源激发水下声场的实验。采用汽笛作为空气声源,海底放置水听器作为接收,在实验过程中,发射船由距离水听器2.4 km处行驶至9.8 km。本文对该次实验数据进行分析,获得了收发距离远达9.8 km、频率分别为128 Hz和256 Hz的声传播损失曲线,该曲线随传播距离变化存在清晰的震荡结构.利用波数积分方法计算实验环境下的水下声场理论值,并对获得的声场传播特性进行了较好的物理解释。     

Source levels of clicks from free-ranging white-beaked dolphins (Lagenorhynchus albirostris Gray 1846) recorded in Icelandic waters

This study reports the source levels of clicks recorded from free-ranging white-beaked dolphins (Lagenorhynchus albirostris Gray 1846). A four-hydrophone array was used to obtain sound recordings. The hydrophone signals were digitized on-line and stored in a portable computer. An underwater video camera was used to visualize dolphins to help identify on-axis recordings. The range to a dolphin was calculated from differences in arrival times of clicks at the four hydrophones, allowing for calculations of source levels. Source levels in a single click train varied from 194 to 211 dB peak-to-peak (p-p) re: 1 microPa. The source levels varied linearly with the log of range. The maximum source levels recorded were 219 dB (p-p) re: 1 microPa.     

An echolocation visualization and interface system for dolphin research

The present study describes the development and testing of a tool for dolphin research. This tool was able to visualize the dolphin echolocation signals as well as function as an acoustically operated "touch screen." The system consisted of a matrix of hydrophones attached to a semitransparent screen, which was lowered in front of an underwater acrylic panel in a dolphin pool. When a dolphin aimed its sonar beam at the screen, the hydrophones measured the received sound pressure levels. These hydrophone signals were then transferred to a computer where they were translated into a video image that corresponds to the dynamic sound pressure variations in the sonar beam and the location of the beam axis. There was a continuous projection of the image back onto the hydrophone matrix screen, giving the dolphin an immediate visual feedback to its sonar output. The system offers a whole new experimental methodology in dolphin research and since it is software-based, many different kinds of scientific questions can be addressed. The results were promising and motivate further development of the system and studies of sonar and cognitive abilities of dolphins.     

识别与提取简正波的一种简化方法

本文提出了一种识别与提取浅海简正波的简化方法。海上实验时我们利用同一个水听器,在不同深度接收固定距离上的爆炸信号,实验室回放时爆炸声信号经高频(8kHz)与低频(<0.8kHz)带通滤波,并用高频信号作为低频信号时间采样的起点。理论与实验表明,这样得到的不同接收深度的低频信号相当于一个垂直水听器阵接收的结果。对不同接收深度的低频信号分别按简正波本征函数或按本征函数的符号加权相加,较好地分离出前二号简正波。     

使用深度学习的多通道水下目标识别

为解决低信噪比条件下水下目标识别率低的问题,提出一种适用于多通道水听器阵列的深度学习水下目标识别方法。首先是采用子通道特征级联的方法利用多通道信息;在特征提取方面,采用对信号的不同频率区间进行加权的特征提取器,并对提取的特征进行正则规整;最后采用深度神经网络(Deep Neural Network,DNN)实现目标识别。实验首先在仿真条件下对所提出方法的有效性进行验证,结果表明在-15 dB信噪比条件下的五目标识别任务中,使用多通道级联特征的深度神经网络的识别正确率达到96.7%,显著高于基于支持向量机(Support Vector Machine,SVM)的方法。在后续的湖上试验中,深度神经网络的平均正确率达到96.0%,进一步验证了所提出方法的有效性。     

Long distance passive localization of vocalizing sei whales using an acoustic normal mode approach

During a 2 day period in mid-September 2006, more than 200, unconfirmed but identifiable, sei whale (Balaenoptera borealis) calls were collected as incidental data during a multidisciplinary oceanography and acoustics experiment on the shelf off New Jersey. Using a combined vertical and horizontal acoustic receiving array, sei whale movements were tracked over long distances (up to tens of kilometers) using a normal mode back propagation technique. This approach uses low-frequency, broadband passive sei whale call receptions from a single-station, two-dimensional hydrophone array to perform long distance localization and tracking by exploiting the dispersive nature of propagating normal modes in a shallow water environment. The back propagation approach is examined for accuracy and application to tracking the sei whale vocalizations identified in the vertical and horizontal array signals. This passive whale tracking, combined with the intensive oceanography measurements performed during the experiment, was also used to examine sei whale movements in relation to oceanographic features observed in this region.     

基于光纤水听器的声场水平相关性研究

中科院声学研究所声场声信息国家重点实验室与长沙国防科技大学合作研制出了一条32阵元的光纤水听器阵,并于2002年8月在渤海进行了一次海上实验。本文利用光纤水听器阵获得的气枪声信号进行了声场的水平相关性分析,并与理论结果进行了比较,实验和理论结果基本符合。     

Passive acoustic detection and localization of whales: effects of shipping noise in Saguenay-St. Lawrence Marine Park

The performance of large-aperture hydrophone arrays to detect and localize blue and fin whales' 15-85 Hz signature vocalizations under ocean noise conditions was assessed through simulations from a normal mode propagation model combined to noise statistics from 15 960 h of recordings in Saguenay-St. Lawrence Marine Park. The probability density functions of 2482 summer noise level estimates in the call bands were used to attach a probability of detection/masking to the simulated call levels as a function of whale depth and range for typical environmental conditions. Results indicate that call detection was modulated by the calling depth relative to the sound channel axis and by modal constructive and destructive interferences with range. Masking of loud infrasounds could reach 40% at 30 km for a receiver at the optimal depth. The 30 dB weaker blue whale D-call were subject to severe masking. Mapping the percentages of detection and localization allowed assessing the performance of a six-hydrophone array under mean- and low-noise conditions. This approach is helpful for optimizing hydrophone configuration in implementing passive acoustic monitoring arrays and building their detection function for whale density assessment, as an alternative to or in combination with the traditional undersampling visual methods.     

基于多径分集的啁啾扩频正交频分复用水声通信系统

针对传统正交频分复用系统在具有复杂多途和深度频率选择性衰落特点的水声信道中性能下降的问题,提出了啁啾扩频正交频分复用水声通信方案.该方案对原信息码扩频,子载波变为相同调频斜率、不同中心频率、频带相互重叠的正交啁啾信号.经过水声相干多途信道后在接收端解扩,使多个途径信号在频域上拓展,多径信号彼此分离.结合虚拟时间反转镜技术,聚焦多途信道能量,完成信道多径分集接收,不仅可以抑制频率选择性衰落的影响,还充分利用信道多径分集增益提高系统性能.通过仿真研究和湖试验证,表明该方案具有较好的有效性和可靠性.     

基于频散特征的单水听器模式特征提取及距离深度估计研究

针对浅海环境中低频宽带水声脉冲信号, 研究基于频散特征结合时频分析的单水听器距离和深度估计方法. 以简正波理论为依据, 将单水听器上的接收信号表示成一系列传播模式之和的形式, 分析了经典波导环境下的频散现象, 采用自适应径向高斯核函数的时频分析方法来表征接收信号的频散特征. 为提高时频分辨率, 采用自适应径向高斯核函数的时频分布来提取频散关系曲线中传播模式的到达时间差, 利用模式的到达时间差估计声源的距离. 采用多模式联合匹配的方式, 通过二值掩模滤波的时频滤波方法, 提取所需的模式. 通过计算实际提取出的模式能量与预测的模式能量之间的误差, 建立代价函数, 并通过模式能量匹配的方式, 确定声源的深度. 通过对基于Pekeris波导模型的浅海环境进行仿真验证, 结果表明: 自适应径向高斯核函数的时频分析方法能够很好地反映信号本身的频散特征, 具有较高的时频分辨率, 克服了传统短时傅里叶变换时频表征的限制, 使得模式在时频域更加容易辨识和分离; 从测距效果来看, 不同模式组合下的距离估计结果不同, 采用在时频面上具有较高能量的模式, 可得到较为准确的距离估计; 选用高能量的模式所得的距离估计的相对误差小于2%. 在定深方面, 参与联合匹配的模式个数越多, 代价函数的峰值更加地尖锐, 同时具有低的伪峰, 深度估计的性能会进一步有所提升. 该工作对于研究低频水声脉冲信号的分离和提取具有重大意义. 关键词： 频散信道 时频分析 单水听器 定位     

基于单个矢量水听器和子空间旋转的宽容宽带双目标分辨与跟踪方法

理论证明线性信号处理可以分辨两个不完全相关的目标。针对宽带双目标问题,根据矢量水听器信号模型,设计了并行的两路结构相同的处理通道,每个通道先将接收信号做子空间旋转,分离出单个目标,然后用空间匹配滤波方法估计该目标方位。仿真研究表明,该方法适用于任意频谱结构的信号,对目标强度差和初值误差的宽容性好,可以全方位跟踪目标而性能不受目标具体方位影响,并具有时间带宽增益,在相关目标或者低信噪比情况下的性能优于宽带聚焦子空间方法。海试数据验证了该方法的可行性。该方法是一种比较实用的水下双目标分辨与跟踪方法。     

Acoustics and hydrodynamics of a drop impact on a water surface

Hydrodynamic and acoustic processes associated with a drop impact on a water surface were studied experimentally. Acoustic signals were detected underwater (with a hydrophone) and in air (with a microphone), the flow pattern was recorded with a high-speed camera, and the surface perturbation was monitored with a laser detector. The dimensionless parameters of flows (Reynolds, Froude, and Weber numbers) induced by the impact varied with fall height within the ranges of 5000 < Re < 20000, 20 < Fr < 350, and 70 < We < 1000. The sequence of acoustic signals incorporated an impact pulse at the moment of contact between a drop and the surface and a series of acoustic packets attributable to the resonance emission of gas cavities. The top of the impact pulse, which was detected clearly in the entire fall height range, had a complex structure with short high-frequency and longer low-frequency oscillations. The total number and the parameters of emitted acoustic packets depended to a considerable extent on the fall height. The cases of lacking, one-time, and repeated emission of packets were noted in a series of experiments performed at a constant fall height. The analysis of video data showed that the signal variability was induced by considerable differences in the scenarios of water entry of a drop, which assumed an ovoid shape at the end trajectory segment, in the mentioned experiments.     

High-frequency modulated signals of killer whales (Orcinus orca) in the North Pacific

Killer whales in the North Pacific, similar to Atlantic populations, produce high-frequency modulated signals, based on acoustic recordings from ship-based hydrophone arrays and autonomous recorders at multiple locations. The median peak frequency of these signals ranged from 19.6-36.1 kHz and median duration ranged from 50-163 ms. Source levels were 185-193 dB peak-to-peak re: 1 μPa at 1 m. These uniform, repetitive, down-swept signals are similar to bat echolocation signals and possibly could have echolocation functionality. A large geographic range of occurrence suggests that different killer whale ecotypes may utilize these signals.     

单矢量水听器稀疏近似最小方差方位估计算法

针对单矢量水听器海上目标探测问题,利用稀疏近似最小方差(Sparse Asymptotic Minimum Variance,SAMV)算法进行目标方位估计,该算法利用单矢量水听器自身具有阵列流形的特点,将整个扫描空间离散化,目标方位分布于某一离散方向位置上,利用空间信号的稀疏性可提高目标方位估计性能。仿真结果表明,SAMV算法在各信噪比条件下方位估计噪声背景级明显优于常规波束形成(Conventional Beam Forming,CBF)算法和最小方差无失真响应(Minimum Variance Distortionless Response,MVDR)算法,当信噪比大于0dB时,该算法测向结果均方根误差小于2°,且SAMV算法具有更好的空间方位分辨能力。消声水池和海上声学浮标海上试验数据处理结果表明,SAMV算法给出了噪声背景级更低的目标方位历程图,有效验证了SAMV算法对海上目标的探测性能及其有效性。     

Age-related differences in the acoustic characteristics of male leopard seals, Hydrurga leptonyx

During the breeding season, the underwater vocalizations and calling rates of adult male leopard seals are highly stereotyped. In contrast, sub-adult males have more variable acoustic behavior. Although adult males produce only five stereotyped broadcast calls as part of their long-range underwater breeding displays the sub-adults have a greater repertoire including the adult-like broadcast calls, as well as variants of these. Whether this extended repertoire has a social function is unknown due to the paucity of behavioral data for this species. The broadcast calls of the sub-adults are less stereotyped in their acoustic characteristics and they have a more variable calling rate. These age-related differences have major implications for geographic variation studies, where the acoustic behavior of different populations are compared, as well as for acoustic surveying studies, where numbers of calls are used to indicate numbers of individuals present. Sampling regimes which unknowingly include recordings from sub-adult animals will artificially exaggerate differences between populations and numbers of calling animals. The acoustic behavior of sub-adult and adult male leopard seals were significantly different and although this study does not show evidence that these differences reflect vocal learning in the male leopard seal it does suggest that contextual learning may be present.