Search

Ocean noise negatively influences the recording of odontocete echolocation clicks. In this study, a hybrid model based on the convolutional neural network (CNN) and long short-term memory (LSTM) network-called a hybrid CNN-LSTM model-was proposed to denoise echolocation clicks. To learn the model parameters, the echolocation clicks were partially corrupted by adding ocean noise, and the model was trained to recover the original echolocation clicks. It can be difficult to collect large numbers of echolocation clicks free of ambient sea noise for training networks. Data augmentation and transfer learning were employed to address this problem. Based on Gabor functions, simulated echolocation clicks were generated to pre-train the network models, and the parameters of the networks were then fine-tuned using odontocete echolocation clicks. Finally, the performance of the proposed model was evaluated using synthetic data. The experimental results demonstrated the effectiveness of the proposed model for denoising two typical echolocation clicks-namely, narrowband high-frequency and broadband echolocation clicks. The denoising performance of hybrid models with the different number of convolution and LSTM layers was evaluated. Consequently, hybrid models with one convolutional layer and multiple LSTM layers are recommended, which can be adopted for denoising both types of echolocation clicks., (© 2023 Acoustical Society of America.)

Odontocetes have evolved special acoustic structures in the forehead to modulate echolocation and communication signals into directional beams to facilitate feeding and social behaviors. Whistle directivity was addressed for the Indo-Pacific humpback dolphin (Sousa chinensis) by developing numerical models in the current paper. Directivity was first examined at the fundamental frequency 5 kHz, and simulations were then extended to the harmonics of 10, 15, 20, 25, and 30 kHz. At 5 kHz, the -3 dB beam widths in the vertical and horizontal planes were 149.3° and 119.4°, corresponding to the directivity indexes (DIs) of 4.4 and 5.4 dB, respectively. More importantly, we incorporated directivity of the fundamental frequency and harmonics to produce an overall beam, resulting in -3 dB beam widths of 77.2° and 62.9° and DIs of 8.2 and 9.7 dB in the vertical and horizontal planes, respectively. Harmonics can enhance the directivity of fundamental frequency by 3.8 and 4.3 dB, respectively. These results suggested the transmission system can modulate whistles into directional projection, and harmonics can improve DI.

Ocean noise has a negative impact on the acoustic recordings of odontocetes' echolocation clicks. In this study, deep convolutional autoencoders (DCAEs) are presented to denoise the echolocation clicks of the finless porpoise (Neophocaena phocaenoides sunameri). A DCAE consists of an encoder network and a decoder network. The encoder network is composed of convolutional layers and fully connected layers, whereas the decoder network consists of fully connected layers and transposed convolutional layers. The training scheme of the denoising autoencoder was applied to learn the DCAE parameters. In addition, transfer learning was employed to address the difficulty in collecting a large number of echolocation clicks that are free of ambient sea noise. Gabor functions were used to generate simulated clicks to pretrain the DCAEs; subsequently, the parameters of the DCAEs were fine-tuned using the echolocation clicks of the finless porpoise. The experimental results showed that a DCAE pretrained with simulated clicks achieved better denoising results than a DCAE trained only with echolocation clicks. Moreover, deep fully convolutional autoencoders, which are special DCAEs that do not contain fully connected layers, generally achieved better performance than the DCAEs that contain fully connected layers.

The wave propagation, sound field, and transmission beam pattern of a pygmy sperm whale (Kogia breviceps) were investigated in both the horizontal and vertical planes. Results suggested that the signals obtained at both planes were similarly characterized with a high peak frequency and a relatively narrow bandwidth, close to the ones recorded from live animals. The sound beam measured outside the head in the vertical plane was narrower than that of the horizontal one. Cases with different combinations of air-filled structures in both planes were used to study the respective roles in controlling wave propagation and beam formation. The wave propagations and beam patterns in the horizontal and vertical planes elucidated the important reflection effect of the spermaceti and vocal chambers on sound waves, which was highly significant in forming intensive forward sound beams. The air-filled structures, the forehead soft tissues and skull structures formed wave guides in these two planes for emitted sounds to propagate forward.

A finite element method was used to investigate the temperature influence on sound beams of the Indo-Pacific humpback dolphin. The numerical models of a dolphin, which originated from previous computed tomography (CT) scanning and physical measurement results, were used to investigate sound beam patterns of the dolphin in temperatures from 21 °C to 39 °C, in increments of 2 °C. The -3 dB beam widths across the temperatures ranged from 9.3° to 12.6°, and main beam angle ranged from 4.7° to 7.2° for these temperatures. The subsequent simulation suggested that the dolphin's sound beam patterns, side lobes in particular, were influenced by temperature.

Acoustic properties of odontocete head tissues, including sound velocity, density, and acoustic impedance, are important parameters to understand dynamics of its echolocation. In this paper, acoustic properties of head tissues from a freshly dead short-finned pilot whale (Globicephala macrorhynchus) were reconstructed using computed tomography (CT) and ultrasound. The animal's forehead soft tissues were cut into 188 ordered samples. Sound velocity, density, and acoustic impedance of each sample were either directly measured or calculated by formula, and Hounsfield Unit values (HUs) were obtained from CT scanning. According to relationships between HUs and sound velocity, HUs and density, as well as HUs and acoustic impedance, distributions of acoustic properties in the head were reconstructed. The inner core in the melon with low-sound velocity and low-density is an evidence for its potential function of sound focusing. The increase in acoustic impedance of forehead tissues from inner core to outer layer may be important for the acoustic impedance matching between the outer layer tissue and seawater. In addition, temperature dependence of sound velocity in soft tissues was also examined. The results provide a guide to the simulation of the sound emission of the short-finned pilot whale.

Harbor porpoises (Phocoena phocoena) use narrow band echolocation signals for detecting and locating prey and for spatial orientation. In this study, acoustic impedance values of tissues in the porpoise's head were calculated from computer tomography (CT) scan and the corresponding Hounsfield Units. A two-dimensional finite element model of the acoustic impedance was constructed based on CT scan data to simulate the acoustic propagation through the animal's head. The far field transmission beam pattern in the vertical plane and the waveforms of the receiving points around the forehead were compared with prior measurement results, the simulation results were qualitatively consistent with the measurement results. The role of the main structures in the head such as the air sacs, melon and skull in the acoustic propagation was investigated. The results showed that air sacs and skull are the major components to form the vertical beam. Additionally, both beam patterns and sound pressure of the sound waves through four positions deep inside the melon were demonstrated to show the role of the melon in the biosonar sound propagation processes in the vertical plane.

The relative role of the various structures in the head of the baiji (Lipotes vexillifer) is examined. A finite element approach was applied to numerically simulate the acoustic propagation through a dolphin's head to examine the relative role of the skull, air sacs, and melon in the formation of the biosonar beam in the vertical plane. The beam pattern obtained with the whole head in place is compared with the beam pattern when the air sac is removed and the other structures (skull and melon) are in place, with only the skull removed, and finally with only the melon removed. The beam pattern with the air sacs and skull intact and the melon removed closely resembled the beam pattern for the complete head, suggesting that the melon has a minor role in the formation of the beam. The beam pattern for the other two cases had very little resemblance to the beam pattern for the whole head. The air sacs seem to have a role of directing propagation of the signal toward the front and the skull prevents the sound propagating below the rostrum. The beam patterns along with a correlation analysis showed that the melon had only a slight influence on the shape and direction of the beam. The resultant beam exiting the head of the dolphin is the result of complex reflection processes within the head of the animal.

Through numerically solving the appropriate wave equations, propagation of biosonar signals in a Chinese river dolphin (baiji) was studied. The interfacial waves along the rostrum-tissue interfaces, including both compressional (longitudinal) and shear (transverse) waves in the solid rostrum through fluid-solid coupling were examined. The baiji's rostrum was found to effect acoustic beam formation not only as an interfacial wave generator but also as a sound reflector. The wave propagation patterns in the solid rostrum were found to significantly change the wave movement through the bone. Vibrations in the rostrum, expressed in solid displacement, initially increased but eventually decreased from posterior to anterior sides, indicating a complex physical process. Furthermore, the comparisons among seven cases, including the combination of (1) the rostrum, melon, and air sacs; (2) rostrum-air sacs; (3) rostrum-melon; (4) only rostrum; (5) air sacs-melon; (6) only air sacs; and (7) only melon revealed that the cases including the rostrum were better able to approach the complete system by inducing rostrum-tissue interfacial waves and reducing the differences in main beam angle and -3 dB beam width. The interfacial waves in the rostrum were considered complementary with reflection to determine the obbligato role of the rostrum in the baiji's biosonar emission. The far-field beams formed from complete fluid-solid models and non-fluid-solid models were compared to reveal the effects brought by the consideration of shear waves of the solid structures of the baiji. The results may provide useful information for further understanding the role of the rostrum in this odontocete species.