The methods based on Discrete Fourier Transform (DFT) are the mainstream approaches for frequency estimation of signals in power systems. However, they exhibit unfavorable long response times when confronted with signals experiencing abrupt transients, such as amplitude or phase step changes. To surmount this challenge, a novel methodology leveraging the DFT has been designed to estimate power system signals with accuracy and responsiveness under abrupt transient conditions. The method first constructs the correlation between DFT bins and each parameter. The relationship is then harnessed to derive an unbiased estimator for sine-wave with a known step position. Afterward, we introduce a step position estimation procedure that guarantees the robustness of the estimator when dealing with abrupt transients. As a result, the proposed method achieves zero response time when confronted with arbitrary abrupt transients without loss of accuracy. The effectiveness and responsiveness of our method are evaluated through simulations that adhere to the stringent requirements of P-class phasor measurement units.

Convolutional neural network (CNN) is widely used in acoustic scene classification (ASC) tasks. In most cases, local convolution is utilized to gather time-frequency information between spectrum nodes. It is challenging to adequately express the non-local link between frequency domains in a finite convolution region. In this paper, we propose a dual-path convolutional neural network based on band interaction block (DCNN-bi) for ASC, with mel-spectrogram as the model’s input. We build two parallel CNN paths to learn the high-frequency and low-frequency components of the input feature. Additionally, we have created three band interaction blocks (bi-blocks) to explore the pertinent nodes between various frequency bands, which are connected between two paths. Combining the time-frequency information from two paths, the bi-blocks with three distinct designs acquire non-local information and send it back to the respective paths. The experimental results indicate that the utilization of the bi-block has the potential to improve the initial performance of the CNN substantially. Specifically, when applied to the DCASE 2018 and DCASE 2020 datasets, the CNN exhibited performance improvements of 1.79% and 3.06%, respectively. © 2024 The Institute of Electronics, Information and Communication Engineers.

Purpose: The purpose of this study was to determine the speech recognition equivalence of Mandarin Bamford-Kowal-Bench (BKB) sentence lists with adults and children with normal hearing. Method: A total of 32 lists, each of nine sentences, were compiled from a corpus of BKB-like sentences with paired babble in Mandarin. Interlist equivalence, critical differences, and sensitivity of performance to signal-to-noise ratio (SNR) were examined. Experiment 1 included 64 native Mandarin-speaking adults with normal hearing. Experiment 2 included 54 native Mandarin-speaking children with normal hearing aged 4-6 years. Results: Among the 32 sentence lists, 28 lists were confirmed to be equivalent in adults, with a mean SNR required for 50% correct (SNR50) of -5.9 +/- 0.1 dB, a mean slope of 22.3%/dB +/- 1.5%/dB, and a grand 95% critical difference subsequently calculated as 27.2% for score. From the 28 equivalent lists, 27 lists were selected and observed to be equivalent in children, with a mean SNR50 threshold of -2.0 +/- 0.2 dB, a mean slope of 15.8%/dB +/- 1.1%/dB, and a grand 95% critical difference of 24.6% for score. Conclusions: The Mandarin BKB sentences in babble noise test offers an opportunity for clinicians and researchers to assess speech understanding in adults and preschool children in an efficient manner. For comparisons of performance in different test conditions, 28 equivalent lists are available for adults and 27 equivalent lists for preschool children. The 95% critical difference values can be used for total percentage correct or SNR for 50% performance. Future work will examine the clinical utility for school-age children and children who are deaf and hard of hearing.

The performance of the speech enhancement (SE) algorithm will decrease rapidly in the presence of interference, especially competing or interfering speech. In this article, an improved real-time implementation of the transfer function generalized sidelobe canceller(TF-GSC) method based on distributed dual-microphone is proposed for interference suppression in the specific direction. In our method, we first derive an improved TF-GSC method based on a primary microphone and a secondary microphone which is abbreviated as GSC-PS in the following. GSC-PS estimates the desired signal by the dual-microphone structure based on estimation of time delay of arrival and calculation of the transfer functions. After that, we propose a new adaptive interference canceller based on the multichannel speech presence probability (MC-SPP) and the output gate unit. The calculated MC-SPP is applied to the step size adjustment and cost function modification of the adaptive interference canceller, while the output gate is designed based on the normalized posterior signal-to-interference ratio difference, which is sensitive to the direction of signal sources. The simulation results show that the proposed GSC-PS algorithm outperforms the current mainstream single-channel and multi-channel SE algorithms in suppressing interference and causes less damage to the quality of the target speech. In addition, experimental results confirm the usability of the proposed algorithm in real world acoustic environment with multiple sources of noises.

Proteomics technology reveals the marker proteins, potential pathogenesis, and intervention targets after noise-induced hearing loss. To study the differences in cochlea protein expression before and after noise exposure using proteomics to reveal the pathological mechanism of noise-induced hearing loss (NIHL). A guinea pig NIHL model was established to test the ABR thresholds before and after noise exposure. The proteomics technology was used to study the mechanism of differential protein expression in the cochlea by noise stimulation. The average hearing threshold of guinea pigs on the first day after noise exposure was 57.00 ± 6.78 dB Sound pressure level (SPL); the average hearing threshold on the seventh day after noise exposure was 45.83 ± 6.07 dB SPL. The proteomics technology identified 3122 different inner ear proteins, of which six proteins related to the hearing were down-regulation: Tenascin C, Collagen Type XI alpha two chains, Collagen Type II alpha one chain, Thrombospondin 2, Collagen Type XI alpha one chain and Ribosomal protein L38, and are enriched in protein absorption, focal adhesion, and extracellular matrix receptor pathways. Impulse noise can affect the expression of differential proteins through focal adhesion pathways. This data can provide an experimental basis for the research on the prevention and treatment of NIHL. © 2022, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.

Hearing loss is one of the most prevalent chronic health problems worldwide and a common intervention is the wearing of hearing aids. However, the tedious fitting procedures and limited hearing experts pose restrictions for the popularity of hearing aids. This paper introduced a hearing loss classification method based on the insertion gain of hearing aids, which aims to simplify the fitting procedure and achieve a fitting-free effect of the hearing aid, in line with current research trends in key algorithms for fitting-free hearing aids. The proposed method innovatively combines the insertion gain of hearing aids with the covariates of patient’s gender, age, wearing history to form a new set of hearing loss vectors, and then classifies the hearing loss into six categories by unsupervised cluster analysis method. Each category of representative parameters characterizes a typical type of hearing loss, which can be used as the initial parameter to improve the efficiency of hearing aid fitting. Compared with the traditional audiogram classification method AMCLASS (Automated Audiogram Classification System), the proposed classification method reflect the actual hearing loss of hearing impaired patients better. Moreover, the effectiveness of the new classification method was verified by the comparison between the obtained six sets of representative insertion gains and the inferred hearing personalization information. © 2023, The Author(s).

In human-computer interaction, acoustic scene classification (ASC) is one of the relevant research domains. In real life, the recorded audio may include a lot of noise and quiet clips, making it hard for earlier ASC-based research to isolate the crucial scene information in sound. Furthermore, scene information may be scattered across numerous audio frames; hence, selecting scene-related frames is crucial for ASC. In this context, an integrated convolutional neural network with a fusion attention mechanism (ICNN-FA) is proposed for ASC. Firstly, segmented mel-spectrograms as the input of ICNN can assist the model in learning the short-Term time-frequency correlation information. Then, the designed ICNN model is employed to learn these segment-level features. In addition, the proposed global attention layer may gather global information by integrating these segment features. Finally, the developed fusion attention layer is utilized to fuse all segment-level features while the classifier classifies various situations. Experimental findings using ASC datasets from DCASE 2018 and 2019 indicate the efficacy of the suggested method. © 2023 Institute of Electronics, Information and Communication, Engineers, IEICE. All rights reserved.

With the increasing severity of hearing problems caused by an aging population, it has become more urgent to study speech quality evaluation methods for hearing aids to automatically judge the effect of parameter adjustments to solve the problem of lack of audiologists. A new evaluation algorithm, which is based on the multi-task learning strategy that combines the main task of non-intrusive speech quality evaluation with the auxiliary task of quality classification, is proposed. The main task network is com-posed of a bi-directional long short-term memory (BiLSTM) network combined with the self-attention mechanism, and the auxiliary task network is formed by fully connected layers based on batch normal-ization. The proposed algorithm firstly uses a convolutional neural network (CNN) to extract frame-level features. Then, the features are inputted into the main task network and the auxiliary task network to extract utterance-level features, respectively. Finally, the objective scores and quality ratings are obtained by the different mappers. Evaluation experiments of the Hearing Aid Speech Quality Index (HASQI) show that compared with others algorithms, the proposed algorithm effectively improves the speech quality evaluation. And it also shows strong robustness and good generalization performances under different distorted conditions. (c) 2023 Elsevier Ltd. All rights reserved.

Masking-based and spectrum mapping-based methods are the two main algorithms of speech enhancement with deep neural network (DNN). But the mapping-based methods only utilizes the phase of noisy speech, which limits the upper bound of speech enhancement performance. Masking-based methods need to accurately estimate the masking which is still the key problem. Combining the advantages of above two types of methods, this paper proposes the speech enhancement algorithm MM-RDN (masking-mapping residual dense network) based on masking-mapping (MM) and residual dense network (RDN). Using the logarithmic power spectrogram (LPS) of consecutive frames, MM estimates the ideal ratio masking (IRM) matrix of consecutive frames. RDN can make full use of feature maps of all layers. Meanwhile, using the global residual learning to combine the shallow features and deep features, RDN obtains the global dense features from the LPS, thereby improves estimated accuracy of the IRM matrix. Simulations show that the proposed method achieves attractive speech enhancement performance in various acoustic environments. Specifically, in the untrained acoustic test with limited priors, e.g., unmatched signal-to-noise ratio (SNR) and unmatched noise category, MM-RDN can still outperform the existing convolutional recurrent network (CRN) method in the measures of perceptual evaluation of speech quality (PESQ) and other evaluation indexes. It indicates that the proposed algorithm is more generalized in untrained conditions. © 2023 Tech Science Press. All rights reserved.