Speaker recognition is a task of identifying persons from their voices. Recently, deep learning has dramatically revolutionized speaker recognition. However, there is lack of comprehensive reviews on ...the exciting progress. In this paper, we review several major subtasks of speaker recognition, including speaker verification, identification, diarization, and robust speaker recognition, with a focus on deep-learning-based methods. Because the major advantage of deep learning over conventional methods is its representation ability, which is able to produce highly abstract embedding features from utterances, we first pay close attention to deep-learning-based speaker feature extraction, including the inputs, network structures, temporal pooling strategies, and objective functions respectively, which are the fundamental components of many speaker recognition subtasks. Then, we make an overview of speaker diarization, with an emphasis of recent supervised, end-to-end, and online diarization. Finally, we survey robust speaker recognition from the perspectives of domain adaptation and speech enhancement, which are two major approaches of dealing with domain mismatch and noise problems. Popular and recently released corpora are listed at the end of the paper.
Fusing the advantages of multiple acoustic features is important for the robustness of voice activity detection (VAD). Recently, the machine-learning-based VADs have shown a superiority to ...traditional VADs on multiple feature fusion tasks. However, existing machine-learning-based VADs only utilize shallow models, which cannot explore the underlying manifold of the features. In this paper, we propose to fuse multiple features via a deep model, called deep belief network (DBN). DBN is a powerful hierarchical generative model for feature extraction. It can describe highly variant functions and discover the manifold of the features. We take the multiple serially-concatenated features as the input layer of DBN, and then extract a new feature by transferring these features through multiple nonlinear hidden layers. Finally, we predict the class of the new feature by a linear classifier. We further analyze that even a single-hidden-layer-based belief network is as powerful as the state-of-the-art models in the machine-learning-based VADs. In our empirical comparison, ten common features are used for performance analysis. Extensive experimental results on the AURORA2 corpus show that the DBN-based VAD not only outperforms eleven referenced VADs, but also can meet the real-time detection demand of VAD. The results also show that the DBN-based VAD can fuse the advantages of multiple features effectively.
Between the sheets: Sodium‐ion batteries are an attractive, low‐cost alternative to lithium‐ion batteries. Nitrogen‐doped porous carbon sheets are prepared by chemical activation of ...polypyrrole‐functionalized graphene sheets. When using the sheets as anode material in sodium‐ion batteries, their unique compositional and structural features result in high reversible capacity, good cycling stability, and high rate capability.
With the growing demand for solid, portable, and wearable electronics, exploring recyclable and stable charging and cooling techniques is of significance. Fiber-based thermoelectrics, enabling ...sustainable power generation driven by the temperature difference or refrigeration without noise and freon, exhibit great potential for application in advanced electronics. In this work, we review significant advances in fiber-based thermoelectrics, including inorganic fibers, organic fibers, inorganic/organic hybrid fibers, and fiber-based fabrics and devices. The fundamentals, synthesis, characterizations, property evaluation, and applications of thermoelectric fibers are comprehensively discussed with carefully selected cases, and corresponding thermoelectric devices based on these advanced fibers are introduced for both power generation and refrigeration. Furthermore, we point out challenges and future directions toward the development of fiber-based thermoelectrics.
This review comprehensively summarizes the recent progress of fiber-based thermoelectric materials and devices for solid, portable, and wearable electronics.
Owing to the sustainability, environmental friendliness, and structural diversity of biomass‐derived materials, extensive efforts have been devoted to use them as energy storage materials in ...high‐energy rechargeable batteries. A timely and comprehensive review from the structures to mechanisms will significantly widen this research field. Here, it starts with the operation mechanism of batteries, and it aims to summarize the latest advances for biomass‐derived carbon to achieve high‐energy battery materials, including activation carbon methods and the structural classification of biomass‐derived carbon materials from zero dimension, one dimension, two dimension, and three dimension. Each strategy starts with carefully selected examples and then moves to illustrate the underlying transport mechanism of electrons in the structure. In the end, challenges, strategies, and outlooks are pointed out for the future development of biomass‐derived carbon materials. Overall, this review will help researchers choose appropriate strategies to design biomass‐derived carbon materials, thereby promoting the application of biomass materials in battery design.
This review comprehensively summarizes the internal structure of biomass‐derived carbon materials, which aims to provide suitable environment‐friendly and low consumption green materials for high‐performance batteries design.
Display omitted
With the ever-growing development of multifunctional and miniature electronics, the exploring of high-power microwatt-milliwatt self-charging technology is highly essential. Flexible ...thermoelectric materials and devices, utilizing small temperature difference to generate electricity, exhibit great potentials to provide the continuous power supply for wearable and implantable electronics. In this review, we summarize the recent progress of flexible thermoelectric materials, including conducting polymers, organic/inorganic hybrid composites, and fully inorganic materials. The strategies and approaches for enhancing the thermoelectric properties of different flexible materials are detailed overviewed. Besides, we highlight the advanced strategies for the design of mechanical robust flexible thermoelectric devices. In the end, we point out the challenges and outlook for the future development of flexible thermoelectric materials and devices.
Organic tailored materials using various aromatic carbonyl derivative polyimides are synthesized by tuning the alteration of the conjugated backbone. These materials are used as the cathodes for ...high‐power, long‐cycle, and sustainable sodium‐organic batteries.
Keeping mammalian gastrointestinal (GI) tract communities in balance is crucial for host health maintenance. However, our understanding of microbial communities in the GI tract is still very limited. ...In this study, samples taken from the GI tracts of C57BL/6 mice were subjected to 16S rRNA gene sequence-based analysis to examine the characteristic bacterial communities along the mouse GI tract, including those present in the stomach, duodenum, jejunum, ileum, cecum, colon and feces. Further analyses of the 283,234 valid sequences obtained from pyrosequencing revealed that the gastric, duodenal, large intestinal and fecal samples had higher phylogenetic diversity than the jejunum and ileum samples did. The microbial communities found in the small intestine and stomach were different from those seen in the large intestine and fecal samples. A greater proportion of Lactobacillaceae were found in the stomach and small intestine, while a larger proportion of anaerobes such as Bacteroidaceae, Prevotellaceae, Rikenellaceae, Lachnospiraceae, and Ruminococcaceae were found in the large intestine and feces. In addition, inter-mouse variations of microbiota were observed between the large intestinal and fecal samples, which were much smaller than those between the gastric and small intestinal samples. As far as we can ascertain, ours is the first study to systematically characterize bacterial communities from the GI tracts of C57BL/6 mice.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Monaural speech separation is a fundamental problem in robust speech processing. Recently, deep neural network (DNN)-based speech separation methods, which predict either clean speech or an ideal ...time-frequency mask, have demonstrated remarkable performance improvement. However, a single DNN with a given window length does not leverage contextual information sufficiently, and the differences between the two optimization objectives are not well understood. In this paper, we propose a deep ensemble method, named multicontext networks, to address monaural speech separation. The first multicontext network averages the outputs of multiple DNNs whose inputs employ different window lengths. The second multicontext network is a stack of multiple DNNs. Each DNN in a module of the stack takes the concatenation of original acoustic features and expansion of the soft output of the lower module as its input, and predicts the ratio mask of the target speaker; the DNNs in the same module employ different contexts. We have conducted extensive experiments with three speech corpora. The results demonstrate the effectiveness of the proposed method. We have also compared the two optimization objectives systematically and found that predicting the ideal time-frequency mask is more efficient in utilizing clean training speech, while predicting clean speech is less sensitive to SNR variations.
Short-time frequency transform (STFT) is fundamental in speech processing. Because of the difficulty of processing highly unstructured STFT phase, most speech-processing algorithms only operate with ...STFT magnitude, leaving the STFT phase far from explored. However, with the recent development of deep neural network (DNN) based speech processing, e.g., speech enhancement and recognition, phase processing is becoming more important than ever before as a new growing point of DNN-based methods. In this paper, we propose a phase-aware speech enhancement algorithm based on DNN. Specifically, in the training stage, when incorporating phase as a target, our core idea is to transform an unstructured phase spectrogram to its derivative along the time axis, i.e., instantaneous frequency deviation (IFD), which has a similar structure with its corresponding magnitude spectrogram. We further propose to optimize both IFD and magnitude jointly in a multiobjective learning framework. In the test stage, we propose a postprocessing method to recover the phase spectrogram from the estimated IFD. Experimental results demonstrate the effectiveness of the proposed method.