• The Journal of the Acoustical Society of Korea
• /
• v.35 no.2
• /
• pp.118-124
• /
• 2016

In this paper, contribution analysis method using a partial coherence function is dealt with in the case of underwater radiation noise. When performing the contribution analysis using a partial coherence function, it is important to select the order of system input. But in the case of frequency correlated systems, it is very difficult to properly select the order of system input. In order to solve this problem, the contribution analysis is performed by subdividing the area of contribution using multiple coherence function. And the new contribution analysis method is presented by using the relationship between the contribution characteristic matrix and multiple coherence function. In order to validate the new method, calculation is performed about multi-input / single-output model which is composed of sine waves. The result of calculation shows that it is possible to derive the exact contribution values.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.496-502
• /
• 2021

Recently, deep speaker embedding approach has been widely used in text-independent speaker verification, which shows better performance than the traditional i-vector approach. In this work, to improve the deep speaker embedding approach, we propose a novel method called group-based speaker embedding which incorporates group information. We cluster all speakers of the training data into a predefined number of groups in an unsupervised manner, so that a fixed-length group embedding represents the corresponding group. A Group Decision Network (GDN) produces a group weight, and an aggregated group embedding is generated from the weighted sum of the group embeddings and the group weights. Finally, we generate a group-based embedding by adding the aggregated group embedding to the deep speaker embedding. In this way, a speaker embedding can reduce the search space of the speaker identity by incorporating group information, and thereby can flexibly represent a significant number of speakers. We conducted experiments using the VoxCeleb1 database to show that our proposed approach can improve the previous approaches.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.2
• /
• pp.222-230
• /
• 2019

In this paper, the power of the energy harvesting circuit using the PVDF (Polyvinylidene fluoride) piezoelectric sensor transformed by vortex was analyzed. For power analysis, a general bridge diode rectifier circuit and a P-SSHI (Parallel Synchronized Switch Harvesting on Inductor) rectifier circuit with a switching circuit were used. The P-SSHI circuit is a circuit that incorporates a parallel synchronous switch circuit at the input of a general rectifier circuit to improve energy conversion efficiency. In this paper, the output power of general rectifier circuit and P-SSHI rectifier circuit is analyzed and verified through theory and experiment. It was confirmed that the efficiency was increased by 69 % through the experiment using the wind. In addition, a circuit for storing the harvested energy in the supercapacitor was implemented to confirm its applicability as a secondary battery.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.1
• /
• pp.106-113
• /
• 2019

In this paper, we propose an algorithm to classify the received active pulse when the active sonar system is operated as a non-cooperative mode. The proposed algorithm uses CNN (Convolutional Neural Networks) which shows good performance in various fields. As an input of CNN, time frequency analysis data which performs STFT (Short Time Fourier Transform) of the received signal is used. The CNN used in this paper consists of two convolution and pulling layers. We designed a database based neural network and a pulse feature based neural network according to the output layer design. To verify the performance of the algorithm, the data of 3110 CW (Continuous Wave) pulses and LFM (Linear Frequency Modulated) pulses received from the actual ocean were processed to construct training data and test data. As a result of simulation, the database based neural network showed 99.9 % accuracy and the feature based neural network showed about 96 % accuracy when allowing 2 pixel error.

• Journal of IKEEE
• /
• v.24 no.4
• /
• pp.1122-1128
• /
• 2020

Bandwidth Extension refers to restoring and expanding a narrow band signal(NB) that is damaged or damaged in the encoding and decoding process due to the lack of channel capacity or the characteristics of the codec installed in the mobile communication device. It means converting to a wideband signal(WB). Bandwidth extension research mainly focuses on voice signals and converts high bands into frequency domains, such as SBR (Spectral Band Replication) and IGF (Intelligent Gap Filling), and restores disappeared or damaged high bands based on complex feature extraction processes. In this paper, we propose a model that outputs an bandwidth extended signal based on an autoencoder among deep learning models, using the residual connection of one-dimensional convolutional neural networks (CNN), the bandwidth is extended by inputting a time domain signal of a certain length without complicated pre-processing. In addition, it was confirmed that the damaged high band can be restored even by training on a dataset containing various types of sound sources including music that is not limited to the speech.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.1
• /
• pp.10-17
• /
• 2021

The cymbal transducer is a miniaturized version of the Class V flextensional transducer. It has low resonant frequency and high output pressure characteristics compared with its size. However, since it has high quality factor and low energy conversion efficiency as well, it is often used as an array rather than single. When used as an array, a big change in the frequency characteristics occurs in comparison with that of the single transducer due to the interaction between constituent transducers. In this study, we designed a pattern of cymbal array with a view to having broadband characteristics. Three transducers having different center frequencies were designed first. The designed cymbal transducers were used to construct all possible patterns of a 3 × 3 planar array. After analyzing frequency characteristics of these patterns, based on the results, we derived the most effective pattern to achieve a higher fractional bandwidth. The derived array pattern showed an improvement of the fractional bandwidth by 24.9 % in comparison with the reference model.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.3
• /
• pp.213-218
• /
• 2021

It is known that total least-squares method shows better estimation performance than least-squares method when noise is present at the input and output at the same time. When total least squares method is applied to data with time series characteristics, Recursive Total Least Squares (RTS) algorithm has been proposed to improve the real-time performance. However, RTLS has numerical instability in calculating the inverse matrix. In this paper, we propose an algorithm for reducing numerical instability as well as having similar convergence to RTLS. For this algorithm, we propose a new RTLS using Iterative Wiener Filter (IWF). Through the simulation, it is shown that the convergence of the proposed algorithm is similar to that of the RTLS, and the numerical robustness is superior to the RTLS.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.176-182
• /
• 2021

This paper evaluates and compares the performance of the Deep Nerual Network (DNN)-based speech enhancement models according to various loss functions. We used a complex network that can consider the phase information of speech as a baseline model. As the loss function, we consider two types of basic loss functions; the Mean Squared Error (MSE) and the Scale-Invariant Source-to-Noise Ratio (SI-SNR), and two types of perceptual-based loss functions, including the Perceptual Metric for Speech Quality Evaluation (PMSQE) and the Log Mel Spectra (LMS). The performance comparison was performed through objective evaluation and listening tests with outputs obtained using various combinations of the loss functions. Test results show that when a perceptual-based loss function was combined with MSE or SI-SNR, the overall performance is improved, and the perceptual-based loss functions, even exhibiting lower objective scores showed better performance in the listening test.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.1
• /
• pp.57-66
• /
• 2023

In this paper, it was investigated how the amount of energy harvesting will be varied from the FTEH which has inlet area is wider than outer area and attaching cantilever type MFC (Macro Fiber Composite) using by theoretical and experimental approaches. When MFC length increased 50 % vibration displacement also increased 3.5 times. When thickness decreased vibration displacement increased 30.9 times. In underwater tank experiments FTEH with spiral screw, flexible support, vertical direction fabrication cases showed maximum energy harvesting more 5 times than the case of MFC installed horizontally without spiral screws and on rigid supports. When the flow speed of 0.24 m/s FTEH's optimal resistance applied 4,10 kΩ, energy storage in the capacitor was measured 4 ㎼·s during 350 seconds. It was confirmed that the charging energy can be increased by lengthening the capacitor charging time of the large-area MFC installed vertically on the flexible support at high flow speed.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.2
• /
• pp.94-101
• /
• 2023

A deep neural network composed of encoders and decoders, such as U-Net, used for speech enhancement, concatenates the encoder to the decoder through skip-connection. Skip-connection helps reconstruct the enhanced spectrum and complement the lost information. The features of the encoder and the decoder connected by the skip-connection are incompatible with each other. In this paper, for complex-valued spectrum based speech enhancement, Self-Attention (SA) method is applied to skip-connection to transform the feature of encoder to be compatible with the features of decoder. SA is a technique in which when generating an output sequence in a sequence-to-sequence tasks the weighted average of input is used to put attention on subsets of input, showing that noise can be effectively eliminated by being applied in speech enhancement. The three models using encoder and decoder features to apply SA to skip-connection are studied. As experimental results using TIMIT database, the proposed methods show improvements in all evaluation metrics compared to the Deep Complex U-Net (DCUNET) with skip-connection only.