Audio-visual keyword transformer for unconstrained sentence-level keyword spotting

As one of the most effective methods to improve the accuracy and robustness of speech tasks,the audio-visual fusion approach has recently been introduced into the field of Keyword Spotting(KWS).However,existing audio-visual keyword spotting models are limited to detecting isolated words,while keyword spotting for unconstrained speech is still a challenging problem.To this end,an Audio-Visual Keyword Transformer(AVKT)network is proposed to spot keywords in unconstrained video clips.The authors present a transformer classifier with learnable CLS tokens to extract distinctive keyword features from the variable-length audio and visual inputs.The outputs of audio and visual branches are combined in a decision fusion module.As humans can easily notice whether a keyword appears in a sentence or not,our AVKT network can detect whether a video clip with a spoken sentence contains a pre-specified keyword.Moreover,the position of the keyword is localised in the attention map without additional position labels.Exper-imental results on the LRS2-KWS dataset and our newly collected PKU-KWS dataset show that the accuracy of AVKT exceeded 99%in clean scenes and 85%in extremely noisy conditions.The code is available at https://github.com/jialeren/AVKT.

On‐device audio‐visual multi‐person wake word spotting

Audio‐visual wake word spotting is a challenging multi‐modal task that exploits visual information of lip motion patterns to supplement acoustic speech to improve overall detection performance.However,most audio‐visual wake word spotting models are only suitable for simple single‐speaker scenarios and require high computational complexity.Further development is hindered by complex multi‐person scenarios and computational limitations in mobile environments.In this paper,a novel audio‐visual model is proposed for on‐device multi‐person wake word spotting.Firstly,an attention‐based audio‐visual voice activity detection module is presented,which generates an attention score matrix of audio and visual representations to derive active speaker representation.Secondly,the knowledge distillation method is introduced to transfer knowledge from the large model to the on‐device model to control the size of our model.Moreover,a new audio‐visual dataset,PKU‐KWS,is collected for sentence‐level multi‐person wake word spotting.Experimental results on the PKU‐KWS dataset show that this approach outperforms the previous state‐of‐the‐art methods.

Proteomics analysis provides novel biomarkers and therapeutic target candidates in the treatment of the Huang-Pu-Tong-Qiao formula in an AD rat mode

Background:Huang-Pu-Tong-Qiao formula(HPTQ),a traditional Chinese herbal formula,has a variety of pharmacological effects.It has been used to treat Alzheimer’s disease(AD)for decades.This study aimed to screen differentially expressed proteins in the hippocampus of AD model rats treated with HPTQ.Proteomic studies of the effects of HPTQ on AD are key to understanding the therapeutic mechanisms of HPTQ and identifying potential therapeutic targets.Methods:We hence used the isobaric tags for relative and absolute quantification(ITRAQ)approach to investigate the differentially expressed proteins in the hippocampus of AD model rats before and after HPTQ administration and to identify the potential therapeutic target proteins of HPTQ.In this study,the learning and memory abilities of AD rats were examined by the Morris water maze test.After HPTQ administration,the differentially expressed proteins in the hippocampus of AD rats were quantified and analyzed in silico.Furthermore,western blotting was used to verify the expression of related proteins.Results:The Morris water maze results showed that HPTQ could improve the learning and memory ability of AD model rats.The proteomics analysis results showed that 57 proteins were differentially expressed,of which 35 were up-regulated and 22 were down-regulated.Bioinformatics analysis indicated that proteins with altered expression after HPTQ treatment were involved in several biological processes that have the potential to exert neuroprotective effects.These included promoting the translation of ribosomes,improving the deposition of amyloid-beta(Aβ),regulating autophagy,regulating neuronal synaptic function and plasticity,and alleviating oxidative stress.Conclusion:In conclusion,we identified several potential therapeutic target proteins and related mechanistic pathways of HPTQ in the treatment of AD,laying the foundation for further investigation of the therapeutic effects of HPTQ.

Houston16:A stable geodetic reference frame for subsidence and faulting study in the Houston metropolitan area,Texas,U.S.

Houston,Texas,is one of the earliest regions to employ continuous GPS technology for land subsidence monitoring.Currently,there are over 200 permanent GPS stations located in the greater Houston area.The need for a consistent and stable reference frame to precisely interpret local ground deformation has become critically important since the 1990s,because of the significant spreading of subsidence.We have established a stable Houston reference frame,designated Houston16,using publicly available GPS observations(>5 years)from 15 Continuously Operating Reference Stations(CORS)located outside of the greater Houston area.Applications of the reference frame in subsidence and faulting studies are demonstrated in this article.Houston16 is aligned in origin and scale with the International GNSS Reference Frame of 2008(IGS08).A 7-parameter method was employed to transform the GPS-derived positional time series from IGS08 to Houston16.The primary product of this study is the seven parameters for transforming Earth-Centered-Earth-Fixed XYZ coordinates from IGS08 to Houston16.The frame stability of Houston16 is approximately 1 mm/year.Houston16 will be updated every few years to mitigate degradation of the frame’s stability with time and to synchronize with future updates of the IGS reference frame.

Ormosia Rondani(Diptera:Limoniidae:Chioneinae),anewly recorded genus fromShandong,Chinawithdescription of one new species

The family Limoniidae includes a large number of species,including agricultural pests,pollinators,and environmental indicators.Here,the genus Ormosia Rondani,1856,a large genus in the familyLimoniidae,is recorded from Shandong,China for the first time.The species O.(O.)kunyushana Xu,Yang&Kang sp.nov.is described and illustrated as new to science.This species is distinguished fromits congenersmainly by body color and male genitalia.A key to known Chinese Ormosia s.str.species is also provided.

Two new and two known species of the genus Togoperla(Plecoptera:Perlidae)from Guangxi of southern China

Two new species of the genus Togoperla are described from Guangxi,China:T.caii sp.nov.and T.hippata sp.nov.Togoperla perpicta Klapálek,1921 and T.triangulata Du&Chou,1999 from the same region are diagnosed and re-described.The taxonomic relationships are discussed with related congeners.

广西花坪SITE100样地甲虫标本照片数据集

通过在全球定点布设长期样地,定量评估昆虫多样性与环境之间的关系具有重要意义。“SITE100”国际大科学计划在全球选定100个大样地,每个样地统一选用罐诱法、马氏网法和飞行阻隔法三类采集装备,从物种、形态和遗传多样性三个维度探究全球昆虫多样性格局。花坪被选为全球100个大样地之一。本文的数据集为花坪样地2020年生长季期间的鞘翅目收样结果,历时7个月(2020.05.09–2020.11.23),涵盖10个样点,每个样点包括1套飞行阻隔器、1套马氏网和10个埋罐,平均每周访问全部样点1次。研究期间共采集甲虫8,914头,被划分为563个形态种,涵盖57个科。数据集包括四部分,第一部分为各样点每个采集装备每次收样所获得甲虫标本的集体照原图,共计644张,压缩后为照片1.zip;第二部分为标注了形态种编号的甲虫标本的集体照,共计644张,压缩后为照片2.zip;第三部分为甲虫标本数量统计数据,涵盖2020年5月2日至2020年11月23日期间3种采集方式各日期采集甲虫的数量、采集样点的经纬度以及海拔信息,共2个文件,压缩后为数据1.zip;第四部分为甲虫标本高级阶元(科级或亚科级)鉴定信息,共计2个文件,压缩后为数据2.zip。该数据集除了可以用于与其他SITE100样地结果进行联合分析之外,还可用来比较不同栖境中甲虫的类群分布和形态差异,部分类群的背面观图片为后续的几何形态学或者形态学研究提供数据支持。该数据集还是对花坪甲虫多样性的首次定量评估,对于丰富我国生物多样性本底数据和了解我国南方昆虫区系的来源具有重要意义。