Combination of dynamic features with a new mask to optimize neural network speech enhancement

Concerning the problem that the Neural Network speech enhancement algorithm cannot fully represent the nonlinear structure of speech due to feature selection,which leads to speech distortion.This paper proposes the combination of dynamic features with a new mask to optimize neural network speech enhancement.First,three features of noisy speech are extracted and spliced to obtain static features.Then,the first and second difference derivatives are obtained to capture the instantaneous signals of speech and fuse them into dynamic features.The combination of dynamic and static features completes internal complementarity of features and reduced speech distortion.Second,in order to enhance the intelligibility and clarity of speech at the same time,an adaptive mask is proposed,which can adjust the energy ratio of speech and noise as well as the ratio of the traditional mask and the square root mask.The Gammatone channel weight is used to modify the mask value in each channel to simulate the human auditory system and further improve the speech intelligibility.Finally,the simulation of multiple voices under different noise backgrounds shows that compared with different literature algorithms,the algorithm has a higher SNR,subjective speech quality and short-term objective intelligibility,which verifies the effectiveness of the algorithm.     

Tibetan Multi-Dialect Speech and Dialect Identity Recognition

Tibetan language has very limited resource for conventional automatic speech recognition so far. It lacks of enough data, sub-word unit, lexicons and word inventories for some dialects. And speech content recognition and dialect classification have been treated as two independent tasks and modeled respectively in most prior works. But the two tasks are highly correlated. In this paper, we present a multi-task WaveNet model to perform simultaneous Tibetan multi-dialect speech recognition and dialect identification. It avoids processing the pronunciation dictionary and word segmentation for new dialects, while, in the meantime, allows training speech recognition and dialect identification in a single model. The experimental results show our method can simultaneously recognize speech content for different Tibetan dialects and identify the dialect with high accuracy using a unified model. The dialect information used in output for training can improve multi-dialect speech recognition accuracy, and the low-resource dialects got higher speech content recognition rate and dialect classification accuracy by multi-dialect and multi-task recognition model than task-specific models.     

针对多种处理痕迹的数字语音取证算法

现有的数字语音取证研究主要集中于对单一的某种操作进行检测，无法对不相关的操作进行判断。针对该问题，提出了一种能够同时检测经过变调、低通滤波、高通滤波和加噪这四种操作的数字语音取证方法。首先，计算语音的归一化梅尔频率倒谱系数（MFCC）统计矩特征；然后通过多个二分类器对特征进行训练，并组合投票得到多分类器；最后使用该多分类器对待测语音进行分类。在TIMIT以及UME语音库上的实验结果表明，归一化MFCC统计矩特征在库内实验中均达到了97%以上的检测率，且在对MP3压缩鲁棒性测试的实验中，检测率仍能保持在96%以上。     

Effect of the composition and degree of crosslinking on the properties of poly(l‐lactic acid)/crosslinked polyurethane blends

Poly(l ‐lactic acid)/crosslinked polyurethane (PLLA/CPU) blends which were prepared via reactive blending of PLLA with poly(?‐caprolactone) (PCL), glycerol and 4,4′‐methylenediphenyl diisocyanate showed excellent toughness. The effects of the composition of the mixture and degree of crosslinking of CPU on the toughness of the PLLA/CPU blends (80/20 w/w) were studied in detail. Dynamic mechanical analysis and rheological measurements were used to characterize the structure of the in situ formed CPU in the PLLA matrix. A novel netlike phase structure was observed when the average molecular weight of PCL and degree of crosslinking were 1 kDa and 10%, respectively. The impact strength of the blend was enhanced from 2.2 kJ m^?2 for pure PLLA to 62.4 kJ m^?2; meanwhile, the elongation at break was increased to 489.8%. Therefore, the mechanical properties of PLLA/CPU blends can be easily tailored by tuning the composition of the mixture and the degree of crosslinking of CPU. © 2018 Society of Chemical Industry     

旋挖钻机设备油液智能分析与健康监测

为了对直升机盒段件入水载荷进行分析，开展了模型垂直入水冲击试验，同时采用有限元-光滑粒子流体动力学（finite element method-smooth particle hydrodynamics，简称FEM-SPH）耦合数值方法进行仿真分析，计算了不同材料构型和速度下的过载、压力及应变结果。与试验结果进行对比表明，仿真结果与试验值吻合较好，预测的结构失效破坏模式与试验结果基本一致，可为真实直升机入水冲击的结构设计提供参考。     

Flexible Piezoelectric Acoustic Sensors and Machine Learning for Speech Processing

Flexible piezoelectric acoustic sensors have been developed to generate multiple sound signals with high sensitivity, shifting the paradigm of future voice technologies. Speech recognition based on advanced acoustic sensors and optimized machine learning software will play an innovative interface for artificial intelligence (AI) services. Collaboration and novel approaches between both smart sensors and speech algorithms should be attempted to realize a hyperconnected society, which can offer personalized services such as biometric authentication, AI secretaries, and home appliances. Here, representative developments in speech recognition are reviewed in terms of flexible piezoelectric materials, self-powered sensors, machine learning algorithms, and speaker recognition.     

Locally one-dimensional-alternating segment explicit–implicit and locally one-dimensional-alternating segment Crank–Nicolson methods for two-dimension parabolic equations

In the paper two new locally one-dimensional alternating segment schemes for solving the 2-D parabolic problems are developed. The parabolic problems can be solved efficiently over decomposed sub-domains. Two schemes are proved to be unconditionally stable and numerical experiments show the validity of the schemes.     

聚氨酯硬段含量变化对酪素/聚氨酯乳液及胶膜性能的影响

以异佛尔酮二异氰酸酯(IPDI)、聚四氢呋喃醚二醇(PTMEG1000)、二羟甲基丙酸(DMPA)、一缩二乙二醇(DEG)等为主要原料,固定酪素与聚氨酯的比例,通过改变聚氨酯硬段含量,制备出系列酪素/聚氨酯乳液(CWPU)。讨论了聚氨酯硬段含量的变化对CWPU状态、胶膜力学性能和热机械性能的影响。结果表明,随着聚氨酯硬段含量增加,聚氨酯分子之间、聚氨酯与酪素分子之间的氢键作用力增大,CWPU的粒径增大,分布变宽,乳液的粒径主要分布在100 nm左右;拉伸性能先增大后减小,硬段含量为64.61%时的拉伸强度达到41.75 MPa,断裂伸长率为344.75%。     

盾构隧道管片衬砌壳-弹簧计算模型研究

针对现有盾构隧道管片衬砌计算模型存在的问题,对管片接头的力学特性进行了分析。认为管片的接头实际应包括管片接缝和对应手孔间的管片部分,采用点弹簧模拟管片接头力学特性实际上是夸大了端部管片的刚度。提出了基于壳-弹簧模式的管片结构的整体环-纵向非连续计算模型,改进了接头的模拟方法。该模型将管片接缝以及对应手孔间管片部分视为一个受力整体,其宽度约为对应手孔之间的弧长,并提出了该接头抗弯刚度的定义方式。由于该模型考虑了管片接头的整体性,计算得到的内力值比梁-弹簧模型计算结果更加准确,且可反映管片结构纵向的受力特征。     

紫外光固化聚氨酯自愈合涂层的合成和性能研究

目的 制备一种紫外光（UV）固化聚氨酯自愈合涂层,并研究其自愈合性能。 方法 以2.4-甲苯二异氰酸酯（TDI）、聚丙二醇（PPG）、三羟基聚醚多元醇（HSH330）、甲基丙烯酸羟乙酯（HEMA）、1,4-丁二醇（BDO）为原料,采用两步法合成UV固化丙烯酸酯改性聚氨酯涂层（APU）。使用红外光谱对合成涂层的化学结构进行表征,使用激光共聚焦显微镜对涂层的表界面性能进行分析,并使用拉力试验机考察涂层的力学性能、自愈合性能及成膜性能。结果 FTIR光谱结果表明,合成的涂层是丙烯酸酯改性聚氨酯涂层（APU）,810 cm–1处存在来自HEMA的—C=C—双键弯曲振动峰,同时涂层仍呈现出明显的聚氨酯化学结构特性。随着HEMA的引入,涂层硬段质量分数从20%增加至40%的过程中,吸水率从36.8%降至3.1%,弹性模量从0.29 MPa增加至8.69 MPa。自愈合性能测试结果表明,随着HEMA的引入,涂层的自愈合率降低,当APU涂层硬段质量分数超过45%后,涂层失去自愈合能力。结论 制备的APU涂层具有优异的耐水性和致密性,由于HEMA含量的增加限制了软段结构的分子链运动和氢键作用,从而削弱了涂层的自愈合性能。