Recently, direct modeling of raw waveforms using deep neural networks has been widely studied for a number of tasks in audio domains. In speaker verification, however, utilization of raw waveforms is in its preliminary phase, requiring further investigation. In this study, we explore end-to-end deep neural networks that input raw waveforms to improve various aspects: front-end speaker embedding extraction including model architecture, pre-training scheme, additional objective functions, and back-end classification. Adjustment of model architecture using a pre-training scheme can extract speaker embeddings, giving a significant improvement in performance. Additional objective functions simplify the process of extracting speaker embeddings by merging conventional two-phase processes: extracting utterance-level features such as i-vectors or x-vectors and the feature enhancement phase, e.g., linear discriminant analysis. Effective back-end classification models that suit the proposed speaker embedding are also explored. We propose an end-toend system that comprises two deep neural networks, one frontend for utterance-level speaker embedding extraction and the other for back-end classification. Experiments conducted on the VoxCeleb1 dataset demonstrate that the proposed model achieves state-of-the-art performance among systems without data augmentation. The proposed system is also comparable to the state-of-the-art x-vector system that adopts data augmentation.
Multidrug resistance 1 (MDR1) is a gene that expresses P-glycoprotein (P-gp), a drug transporter protein. Genetic polymorphisms of MDR1 can be associated with Sasang constitutions because Sasang constitutional medicine (SCM) prescribes different drugs according to different constitutions. A Questionnaire for Sasang Constitution Classification II (QSCC II) was used to diagnose Sasang constitutions. Two hundred and seven healthy people whose Sasang constitutions had been identified were tested. Genotype analyses, restriction fragment length polymorphism (RFLP) and pyrosequencing were used in MDR1 C1236T, and in MDR1 G2677T/A and C3435T, respectively. Significant differences in MDR1 C1236T genotypes were found between So-yangin and So-eumin. MDR1 G2677T/A genotype also showed significant differences in allele distribution between So-yangin and Tae-eumin. So-yangin and So-eumin showed significant differences in the distribution of both 1236C-2677G-3435C and 1236T-2677G-3435T, haplotypes of MDR1. The genetic polymorphism of the MDR1 gene was thus shown to be an indicator that could distinguish So-yangin from other constitutions.
The insulation in buildings is very important. Insulation used in the building is largely divided into organic and inorganic insulation by its insulation material. Organic insulation materials which are made of Styrofoam or polyurethane are extremely vulnerable to fire. On the other hand, inorganic insulation such as mineral wool and glass wool is very weak with moisture, while it is nonflammable, so that its usage is very limited. Therefore, this study developed moisture resistance applicable to mineral wool and glass wool and measured the thermal conductivity of the samples which are exposed to moisture by exposing the product coated with moisture resistance and without moisture resistance to moisture and evaluated how the moisture affects thermal conductivity by applying this to inorganic insulation.
This study attempted to manufacture a Cu-In coating layer via the cold spray process and to investigate the applicability of the layer as a sputtering target material. In addition, changes made to the microstructure and properties of the layer due to annealing heat treatment were evaluated, compared, and analyzed. To examine the microstructural and property changes made to the Cu-In coating layer and Cu coating layer (comparison material), ICP, XRD, SEM, and other tests were conducted; purity, density, hardness, porosity, and bond-strength were measured. The results showed that coating layers with thickness of 20 mm (Cu) and 810 lm (Cu-In) could be manufactured via cold spraying under optimal process conditions. With the Cu-In coating layer, the pure Cu and intermetallic compounds of Cu 7 In 3 and CuIn 4 were found to exist inside the layer regardless of annealing heat treatment. The preannealing inconsistent microstructure of the layer, whose phases were difficult to distinguish was found to have transformed into one with clearer phase distinction and fine, consistent grains following thermal treatment via a progress of recovery, recrystallization, and grain growth. The porosity and hardness values of the coating layers were 1.4% and 133.9 HV, respectively, for Cu and 3.54% and 476.6 HV, respectively, for Cu-In. The values of the Cu-In layer were higher than those of the Cu layer in terms of porosity and hardness, which declined drastically after annealing. With the porosity of the Cu-In coating layer in particular, the higher value found during the preannealing stage dropped to 0.36% after heat treatment of 773 K/1 h as the level on a par with pure Cu (0.44%), thus indicating the improved quality of the Cu-In layer. Moreover, the results of the bond-strength measurement performed on the Cu-In coating layer and annealing treated materials revealed the strength to be relatively high for heat treated coating layers. Based on the findings of this study and on the comparison and discussion of the properties that are typically required of the target material, the Cu-In coating layer manufactured via cold spray process and annealing heat treatment can be said to be applicable as sputtering target in the future.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.