- LI Mengyi;LI Jilong;FENG Haihong;
In order to address the interference caused by the time-varying characteristics of underwater acoustic channels in non-cooperative underwater acoustic communication signal recognition and to meet the needs of low-power deployment, deep learning recognition requires lightweight design to improve recognition accuracy and enhance model generalization ability. A lightweight and efficient recognition model is proposed firstly based on an improved DenseNet structure. By adopting dimension transformation and model compression methods, the model structure and parameters are optimized, reducing the complexity of model inference while ensuring recognition accuracy. Secondly, a multi-modal expression fusion strategy is employed, effectively combining features extracted by different networks,fully utilizing the complementarity of information, thereby significantly improving recognition accuracy. On the simulated dataset, the fusion network achieves a recognition rate of over 94.65% at a signal-to-noise ratio of-6 dB and 98.03% at 0 dB. On the real measured dataset, the accuracy of the base network after transfer learning reaches 98.05%. Lake test results validate the effectiveness of the proposed method.
2025年04期 v.44 457-473页 [查看摘要][在线阅读][下载 1787K] [阅读次数：9 ] |[下载次数：9 ] |[引用频次：0 ] - ZHAO Jixiang;QIN Zhiliang;MA Benjun;LAN Wenjian;YAN Xinyi;ZHENG Yi;
To address the challenges in deep learning-based underwater acoustic source range estimation caused by high-dimensional discrete labels and scarce intra-class samples that constrain model feature learning, this study proposes a siamese neural network-based feature extraction method for target range estimation. First, a dataset containing distancelabeled positive/negative sample pairs was constructed using simulated data, followed by the design and training of a siamese neural network to extract range-discriminative features. Subsequently, siamese neural network feature extraction-based convolutional neural network(S-CNN) and siamese neural network feature extraction-based residual neural network(S-ResNet) were developed through transfer learning strategy. Simulation results demonstrate the method's effectiveness in enhancing range-sensitive feature representation:S-CNN/S-ResNet outperformed baseline models without feature extraction(NS-CNN/NSResNet), with S-ResNet exhibiting superior robustness across varying training sample sizes,signal-to-noise ratios, and environmental uncertainties. The SWellEX-96 experiment validation confirmed the significant advantages of the proposed method over conventional matched-field localization techniques. Notably, the S-ResNet achieved 10% higher confidence probability and 2% lower mean percentage error compared to the S-CNN.
2025年04期 v.44 474-498页 [查看摘要][在线阅读][下载 3796K] [阅读次数：9 ] |[下载次数：5 ] |[引用频次：0 ] - LUO Yuwei;HUO Guanying;CHENG Zhen;
Due to the high cost of data collection and limited experimental conditions,sonar images are often scarce and of poor quality, which hinders effective feature learning and limits the performance of existing detection methods. To address this, we propose an improved YOLO model, i.e. Swin transformer-cascaded group attention YOLO(STCYOLO), for sonar image target detection, which integrates diffusion-based sample generation with a Swin transformer and cascaded group attention(CGA) mechanism. First, we fine-tune stable diffusion via LoRA and incorporate semantic features from the bootstrapping language-image pre-training text model to generate high-quality and diverse sonar images for dataset expansion. Then, we introduce Swin transformer into the YOLOv8 backbone to enhance multi-scale feature extraction for small targets, while integrating the CGA mechanism into the C2f module to improve small object perception. Additionally,the skewed intersection-over-union(SIoU) loss function is utilized to better adapt to the complexities of underwater environments. Experimental results indicate that the trained generative model is capable of producing diverse and realistic samples even in data-scarce scenarios. Compared to the original YOLOv8 model, the enhanced STC-YOLO model exhibits a 5% increase in detection accuracy and a 12.6% improvement in mean average precision, achieving high-precision detection of small underwater targets.
2025年04期 v.44 499-526页 [查看摘要][在线阅读][下载 3046K] [阅读次数：8 ] |[下载次数：6 ] |[引用频次：0 ] - FENG Zhishu;ZHANG Guangpu;ZOU Nan;FU Jin;
In complex underwater acoustic environments, factors such as low signal-to-noise ratio, colored noise, coherent sources, and limited snapshots often significantly degrade the performance of conventional source number estimation methods. Nevertheless,the orthogonal matching pursuit method can still estimate support vectors with considerable accuracy. To robustly estimate the number of sources, this paper constructs two subspaces using the orthogonal matching pursuit method. The two subspaces are generated by the support vectors iteratively solved by the orthogonal matching pursuit method and the eigenvector corresponding to the maximum eigenvalue of the residual. Finally, the two subspaces are used for signal subspace matching criteria to estimate the number of sources.Simulations show that, compared to existing source number estimation methods, the proposed method requires a lower signal-to-noise ratio. It performs better under conditions with a small number of snapshots and is also insensitive to coherent sources and colored noise. The lake trial data processing result shows the method can effectively estimate the number of fixed and moving underwater acoustic targets.
2025年04期 v.44 527-550页 [查看摘要][在线阅读][下载 1105K] [阅读次数：7 ] |[下载次数：4 ] |[引用频次：0 ] - XIE Zhiyuan;WANG Rong;BAI Zonglong;ZHANG Zhijuan;
In order to solve the acoustic imaging problem of two-dimensional block sparse sound sources, a sparse Bayesian learning algorithm with coupling prior hyperparameter is proposed. For two-dimensional block sparse sound source, a Bayesian hierarchical model is established by using parametric coupling method. The sparsity of the sound source within the block is controlled by utilizing coupling constraints, which encourages a block sparse solution. The expectation maximization(EM) algorithm is used to update the hyperparameters iteratively to obtain the sound pressure distribution of the target plane and achieve acoustic imaging. Numerical simulation experiments of sparse sound sources with different structures are designed to compare the performance of the proposed algorithm with existing algorithms. The effects of model parameters, adjacent regions, sound source size and signal-to-noise ratio on the performance of the algorithm are analyzed and acoustic imaging experiments are conducted. The simulation and experimental results show that the proposed method achieves high accuracy performance in the acoustic imaging,exhibits superior performance in sparse sound sources with different structural blocks, and can effectively solve the acoustic imaging problem of two-dimensional sparse sound sources,verifying the effectiveness of this method in practical applications.
2025年04期 v.44 551-572页 [查看摘要][在线阅读][下载 1259K] [阅读次数：8 ] |[下载次数：4 ] |[引用频次：0 ] - LE Ye;HUANG Qingyun;WAN Zhijian;HUANG Jian;CHEN Zhitai;HAN Jianwei;
Surface acoustic wave atomization offers advantages such as concentrated energy, compact size, and low power consumption. However, its efficiency is constrained by instability mechanisms, device frequency, and droplet size. This study focuses on capillary wave disturbances on the surface of micro-droplets during the atomization process.Based on thin-film lubrication theory, the evolution of the thin-film interface under surface acoustic wave excitation is derived, and the characteristics of capillary waves are revealed through numerical simulations. Experimental observations of liquid surface fluctuations validate the frequency of the capillary wave, and the influence of device frequency is analyzed. The results demonstrate that precise control of device frequency and optimization of capillary wave characteristics enable the desired control of aerosol particle size, thereby enhancing atomization efficiency.
2025年04期 v.44 573-590页 [查看摘要][在线阅读][下载 1199K] [阅读次数：8 ] |[下载次数：0 ] |[引用频次：0 ] - CAO Huiyuan;WANG Yuebing;GUO Zhongying;YU Yaping;ZHAO Wei;
The selection of absorbing materials is essential for accurately measuring the acoustic field of focused ultrasound transducers using infrared thermography. This study is based on the physical properties of plastic absorbing films that efficiently absorb sound energy and rapidly convert it to heat. A three-layer medium simulation model based on the principle of sound intensity superposition is established, and a theoretical linear relationship between temperature rise and acoustic intensity is derived within the linear sound field model. In addition, an experimental setup for infrared sound field measurement is developed and the results are validated using the hydrophone scanning method. The results show that the difference in the-6 dB width between the infrared measurements at a power of 30 W for 0.2 s and the acoustic field scanning results is within 0.34%. Under low-power conditions, as the output power increases, the focal temperature rises and the-6 dB width of the focal plane gradually decreases. Over time, the-6 dB width first decreases and then increases, with better matching observed at 0.2 s compared to later stages.
2025年04期 v.44 591-604页 [查看摘要][在线阅读][下载 1118K] [阅读次数：8 ] |[下载次数：7 ] |[引用频次：0 ] - DING Wenjie;XIE Bosun;
Traditional analysis on stereophonic and multichannel sound is based on localization theorem of virtual source, which serves as the basis for designing loudspeaker configuration and signal mixing. However, this traditional theorem is unable to fully explain some psychoacoustic phenomena in reproduction. In present work, a method of analyzing the control of sound field modes in stereophonic and multichannel sound reproduction is proposed. Using Fourier-Bessel expansion, the reproduced sound field is decomposed into a combination of multi-pole modes with various orders. A result consistent with that of traditional virtual source localization theorem is obtained by matching the reproduced modes with that of target sound field up to the first order. By analyzing the contributions of the power of loudspeaker signals to desired and undesired sound field modes, a cost function for desired sound field modes is introduced and serves as a physical criterion of sound field control. Analyses on examples of stereophonic and 5.1-channel sound indicate that an excessive span between a pair adjacent loudspeakers and out-of phase signals for creating virtual source outside the loudspeaker span increase the cost of sound field control, resulting in directional distortion or instability of virtual source. The analyses are validated by psychoacoustic experiments.
2025年04期 v.44 605-636页 [查看摘要][在线阅读][下载 1050K] [阅读次数：7 ] |[下载次数：0 ] |[引用频次：0 ] - WANG Yewei;YU Guangzheng;
The boundary element method, as a numerical approach based on integral equations, has been widely applied in the simulation of head-related transfer functions(HRTFs). However, its preprocessing still requires manual mesh construction, which not only consumes human effort but also compromises the reproducibility of the simulation results. In this study, a representative head mesh template was selected based on morphological correlations of human head shapes, enabling a series of automated preprocessing steps including coordinate alignment, mesh refinement, and source placement at the ear canal entrance, to simplify the computational model and enhance both model processing and HRTF calculation efficiency. Results obtained from the template-based preprocessing demonstrate that, compared with manual processing, the model mesh exhibits a consistent vertex sequence, with rotational and translational errors of the model coordinate system within approximately 3° and 3 mm, respectively, and discrepancies in ear canal source positions not exceeding 2.5 mm. The entire modeling procedure is fully automated by computer,and its time consumption is negligible compared with manual processing. Further analysis indicates that the template-based computational model reduces the calculation time of full-sphere HRTFs by approximately 55%, while maintaining spectral errors within 1.0 dB below 16 kHz, and interaural time difference deviations confined to the lateral directions and less than 10 μs. Finally, validation with a virtual auditory localization model confirmed that no significant differences exist in the localization outcomes of HRTF derived from the two modeling methods. These findings demonstrate that the proposed modeling approach significantly improves the efficiency of HRTF numerical computation without sacrificing accuracy, while optimizing and standardizing the preprocessing workflow of computational models, making it suitable for large-scale personalized HRTF simulations.
2025年04期 v.44 637-655页 [查看摘要][在线阅读][下载 1001K] [阅读次数：8 ] |[下载次数：7 ] |[引用频次：0 ] - YAN Duowen;JIAN Zhihua;CAI Yi;
In response to the issues of low robustness in anonymization and insufficient availability of anonymized speech in downstream tasks, a speaker anonymization method based on adversarial sample generation is proposed. The Adam algorithm is applied to iteratively generate adversarial samples, and speaker features are modified by using these samples to alter the corresponding speaker classification results, thereby achieving speech anonymization. Experimental results demonstrate that, compared with the B1 baseline method of the Voice Privacy Challenge 2024, the method's equal error rate in speaker recognition under semi-informed attacks is improved from 7.64% to 26.30%, greatly enhancing the robustness of the anonymization process. Compared with the B6 baseline method, the method's word error rate in speech recognition is reduced from 9.39% to 4.25%, and, compared with the SOTA system S1, the accuracy in emotion recognition is improved from37.84% to 40.18%, effectively protecting the availability of anonymized speech in downstream tasks. In this approach, adversarial perturbations are used to generate adversarial samples, the speaker classification results of the voice are significantly altered, the original speaker's identity information is concealed, and the robustness of anonymization is enhanced. At the same time, only relatively minor changes are made to the speaker features,thus preserving the data's availability in downstream tasks.
2025年04期 v.44 656-673页 [查看摘要][在线阅读][下载 843K] [阅读次数：7 ] |[下载次数：3 ] |[引用频次：0 ] -
<正>43.05. History; 43.10. General; 43.15. Standards; 43.20. General linear acoustics; 43.25. Nonlinear acoustics; 43.28. Aeroacoustics and atmospheric sound; 43.30. Underwater sound; 43.35. Ultrasonics,quantum acoustics, and physical effects of sound; 43.38. Transduction, acoustical devices for the generation and reproduction of sound; 43.40. Structural acoustics and vibration; 43.50. Noise: its effects and control; 43.55. Architectural acoustics; 43.58. Acoustical measurements and instrumentation;43.60. Acoustic signal processing; 43.64. Physiological acoustics; 43.66. Psychological acoustics; 43.70.Speech production; 43.71. Speech perception; 43.72. Speech processing and communication systems;43.75. Music and musical instruments; 43.80. Bioacoustics; 43.90. Other topics in acoustics.
2025年04期 v.44 674-677页 [查看摘要][在线阅读][下载 113K] [阅读次数：7 ] |[下载次数：2 ] |[引用频次：0 ] <正>~~
2025年04期 v.44 678-681页 [查看摘要][在线阅读][下载 88K] [阅读次数：7 ] |[下载次数：1 ] |[引用频次：0 ] -
<正>ACTA ACUSTICA(声学学报, in Chinese) and Chinese Journal of Acoustics(in English) are the publications of the Institute of Acoustics, Chinese Academy of Sciences,edited by the Acoustical Society of China and the Editorial Board of Chinese Journal of Acoustics, to publish original works in all branches of acoustics, and to promote the international understanding and collaboration in the field of acoustics.
2025年04期 v.44 682页 [查看摘要][在线阅读][下载 32K] [阅读次数：7 ] |[下载次数：1 ] |[引用频次：0 ] <正>1 Chinese Journal of Acoustics is the publication of the Acoustical Society of China and the Institute of Acoustics, Chinese Academy of Sciences, to promote the international understanding and collaboration in the field of acoustics. It publishes original works in all branches of acoustics in China. Contributions from invited guests and friends overseas are also solicited.
2025年04期 v.44 683页 [查看摘要][在线阅读][下载 65K] [阅读次数：7 ] |[下载次数：1 ] |[引用频次：0 ]