A Robust Conformer-Based Speech Recognition Model for Mandarin Air Traffic Control

This study aims to address the deviation in downstream tasks caused by inaccurate recognition results when applying Automatic Speech Recognition(ASR)technology in the Air Traffic Control(ATC)field.This paper presents a novel cascaded model architecture,namely Conformer-CTC/Attention-T5(CCAT),to build a highly accurate and robust ATC speech recognition model.To tackle the challenges posed by noise and fast speech rate in ATC,the Conformer model is employed to extract robust and discriminative speech representations from raw waveforms.On the decoding side,the Attention mechanism is integrated to facilitate precise alignment between input features and output characters.The Text-To-Text Transfer Transformer(T5)language model is also introduced to handle particular pronunciations and code-mixing issues,providing more accurate and concise textual output for downstream tasks.To enhance the model’s robustness,transfer learning and data augmentation techniques are utilized in the training strategy.The model’s performance is optimized by performing hyperparameter tunings,such as adjusting the number of attention heads,encoder layers,and the weights of the loss function.The experimental results demonstrate the significant contributions of data augmentation,hyperparameter tuning,and error correction models to the overall model performance.On the Our ATC Corpus dataset,the proposed model achieves a Character Error Rate(CER)of 3.44%,representing a 3.64%improvement compared to the baseline model.Moreover,the effectiveness of the proposed model is validated on two publicly available datasets.On the AISHELL-1 dataset,the CCAT model achieves a CER of 3.42%,showcasing a 1.23%improvement over the baseline model.Similarly,on the LibriSpeech dataset,the CCAT model achieves a Word Error Rate(WER)of 5.27%,demonstrating a performance improvement of 7.67%compared to the baseline model.Additionally,this paper proposes an evaluation criterion for assessing the robustness of ATC speech recognition systems.In robustness evaluation experiments based on this criterion,the proposed model demonstrates a performance improvement of 22%compared to the baseline model.

Effects of Addition of Amino Acids or Soybean Phospholipid in Diet on Slaughter Performance and Meat Quality of Pigs

[Objectives]The study aims to discuss the effects of addition of arginine and glutamic acid or soybean phospholipid,vitamin E and yeast selenium in diet on the slaughter performance and meat quality of long(white)×large(York)binary hybrid pigs.[Methods]27 long×large castrated hybrid boars with the body weight of(54.4±0.15)kg were randomly divided into 3 groups,with 3 replicates per group and 3 pigs per replicate.Group A was the control group,in which the pigs were fed basal diet;in group B,0.8%arginine and 0.60%glutamate were added to the basal diet;in group C,75 g of soybean phospholipid,20 g of vitamin E and 8 g of yeast selenium were added to every 100 kg of the basal diet.The trial period was 60 d.After the experiment was ended,one test pig with similar body weight was selected from each replicate for slaughter and meat determination.[Results]The average weight gain and eye muscle area of the pigs in group B were significantly higher than those in group C(P<0.05),and also showed an increasing trend compared with group A,but there was no statistically significant difference(P>0.05);there was no significant difference between group B or C and group A in the average weight gain and eye muscle area(P>0.05).There was no significant difference in other slaughter performance between the three groups(P>0.05).Besides,there was also no significant difference in the content of various amino acids,total amino acids and total umami amino acids between the three groups(P>0.05).The inosine content in the longissimus dorsi muscle and muscle cooking loss of binary hybrid pigs in group C were significantly better than those in group B(P<0.05),and also had a tendency to be better than those in group A,but there was no significant difference(P>0.05);there was no significant difference between group B or C and group A in the inosine content and muscle cooking loss of the pigs(P>0.05).In addition,there was no significant difference in other meat traits and chemical composition of the longissimus dorsi muscle between group B or C and group A(P>0.05).[Conclusions]The addition of arginine and glutamic acid or soybean phospholipid,vitamin E and yeast selenium in diet had no significant effect on the growth rate,slaughter performance and meat traits of long×large binary hybrid pigs.

多孔材料微态多尺度计算方法的数值实现与应用

对多孔材料进行有限元建模通常需要极其精细的网格,因而采用直接数值模拟方法有着非常高的建模和计算的成本.传统的一阶计算均匀化框架局限于宏观特征长度远远大于非均质性所在尺度特征长度的问题.为此,本文基于微态多尺度理论,提出多孔材料的微态多尺度计算方法,并应用商业软件MATLAB实现微态多尺度计算均匀化框架.理论上,引入运动场分解,并将宏-微观能量等效通过Hill-Mandel条件自然地联系起来,进而导出宏-微观控制方程.算例结果表明,微态多尺度分析方法拥有较高的计算精度与计算效率.该方法具有良好的通用性,能够为多孔材料的设计和实际工程应用提供有力工具.

A reconfigurable terahertz polarization converter based on metal–graphene hybrid metasurface

A metal-graphene hybrid metasurface polarization converter is designed in this Letter.The unit cell of the hybrid metasurface is composed of a butterfly-shaped structure whose branches are connected by multi-layer graphene sheets.The proposed device can be reconfigured from linear-to-circular polarization to cross-polarization by changing the Fermi energy of graphene.The simulation results show that for three-layer graphene,the device acts as a linear-to-circular polarization converter when EF=0 eV and switches to a cross-polarization converter when EF=0.5 eV.Compared with single-layer graphene,the device with three-layer graphene can maintain the cross-polarization conversion performance under low Fermi energy.Furthermore,two equivalent circuits in the x and y directions are developed to understand the working mechanism of the device.