电话语音回声消除的研究

为了消除电话语音系统中的反射回声,本文通过对电话信道的研究,提出通过估算信道冲击响应,得到提示语音经过信道卷积后的结果,从而抑制回声的算法.与LMS自适应算法相比,它可获得很好的回声抑制和对原始信号的保真效果.回声抵消后,语音识别正确率和准确率分别提高约29%和60%.     

一种任务域无关的语音关键词检测系统

提出了面向语音关键词检测的多尺度声学模型建模框架,基于判决树的自动音素聚类生成了大尺度音素集,利用HMM声学模型训练技术生成了大尺度音素声学上下文相关的背景模型,提高了废料语音的建模精度,还给出了此框架下共享HMM状态的高效搜索空间构造方法,关键词识别准确率平均提高了绝对6.9%;提出了近邻声学上下文准则以及候选关键词在多尺度声学模型上的似然比计算方法并采用FLDA融合,显著提高了声学置信度计算的有效性,系统等错率绝对下降了3.0%。     

含椭圆孔无限大弹性导电薄板在均匀磁场作用下的应力分析

给出一种针对含椭圆孔的导电薄板在电磁力作用下的分析方法.其研究重心是定量计算椭圆孔对电流密度以及板内应力分布在椭圆长轴端点处的集中效应.分析引入椭圆坐标,以助于处理问题的边界条件,并最终给出直角坐标下各分量的对应结果,以便于实际应用.首先求出薄板内电流密度分布, 然后考虑一类简化的计算模型,把应力求解确定为反平面剪切问题, 进而推出应力在板内分布的解析解.并利用所获得的椭圆孔洞结果推断出Ⅲ型裂纹的应力强度因子.     

PESQ测量语音增强算法性能不确定性的研究

对用感知语音质量评价标准（PESQ）测量语音增强算法性能的不确定性进行了研究,给出了不同测试条件下,语音增强算法所得PESQ值的不确定性。通过对结果的分析,比较了轿车噪声条件下几种语音增强算法的性能,并对结果进行了可靠性分析。     

基于PAOLA的丢包补偿算法

VoIP的语音质量受到丢包、延时、时延抖动等因素影响。为提高语音质量,需要进行丢包补偿处理,如采用WSOLA算法。首先介绍了WSOLA算法的原理,然后提出了用PAOLA对其进行优化的方法,最后对恢复语音质量进行了比较。     

Plasma electrolytic liquefaction of cellulosic biomass

In this paper,the rapid liquefaction of a corncob was achieved by plasma electrolysis,providing a new method for cellulosic biomass liquefaction.The liquefaction rate of the corncob was 95％ after 5 min with polyethylene glycol and glycerol as the liquefying agent.The experiments not only showed that H+ ions catalyzed the liquefaction of the corncob,but also that using accelerated H+ ions,which were accelerated by an electric field,could effectively improve the liquefaction efficiency.There was an obvious discharge phenomenon,in which the generated radicals efficiently heated the solution and liquefied the biomass,in the process of plasma electrolytic liquefaction.Finally,the optimum parameters of the corncob liquefaction were obtained by experimentation,and the liquefaction products were analyzed.     

General Steganalysis Method of Compressed Speech Under Different Standards

Analysis-by-synthesis linear predictive coding (AbS-LPC) is widely used in a variety of low-bit-rate speech codecs. Most of the current steganalysis methods for AbS-LPC low-bit-rate compressed speech steganography are specifically designed for a specific coding standard or category of steganography methods, and thus lack generalization capability. In this paper, a general steganalysis method for detecting steganographies in low-bit-rate compressed speech under different standards is proposed. First, the code-element matrices corresponding to different coding standards are concatenated to obtain a synthetic code-element matrix, which will be mapped into an intermediate feature representation by utilizing the pre-trained dictionaries. Then, bidirectional long short-term memory is employed to capture long-term contextual correlations. Finally, a code-element affinity attention mechanism is used to capture the global inter-frame context, and a full connection structure is used to generate the prediction result. Experimental results show that the proposed method is effective and better than the comparison methods for detecting steganographies in cross-standard low-bit-rate compressed speech.     

Design of a compact microstrip tunable dual-band bandpass filter

A novel microstrip tunable dual-band bandpass filter with high attenuation level in the upper stopband region is presented. The proposed structure comprises of T-shaped resonators and the coupled microstrip feeding lines. To reach a compact size, the feeding lines are bended. By altering the physical lengths, the characteristics of BPF can easily tune. So, the operational frequencies of the filter can be adjusted and has been fixed to 2.1 GHz and 2.63 GHz, which are applicative for 3G and 4G systems. The insertion-losses of the first and second pass-bands are only 0.68 dB and 1.08 dB, respectively. The undesired harmonics in the upper stopband, with an excellent attenuation level of −38 dB, are rejected up to 6.65 GHz. The results exhibit high return loss of 19.3 dB and 30.7 dB in the first and second pass-bands. The proposed circuit has been fabricated and tested that the obtained results are well close to the simulations results. The total size of the designed circuit is only 16.4 × 13.4 mm².     

Separation of ultrasonic contributions and energy utilization characteristics of ultrasonic regeneration

The method of applying ultrasound to silica gel regeneration process has been proved to be able to improve regeneration efficiency and reduce regeneration temperature. The average regeneration rate can be increased by 10% and the benefits should owe to the special “vibration effect” and “heating effect” induced by ultrasound. In this article, a dynamic one‐dimensional mathematical model, validated by experiments, has been developed. It is then used to investigate energy utilization efficiency of the ultrasonic regeneration and respective contribution ratio of the two ultrasonic effects. The results show that the contribution ratio of vibration effect dominates and the energy utilization efficiency of silica gel regeneration with ultrasound can be 1% higher. The ultrasonic regeneration is more energy‐saving when the contribution ratio of vibration effect is greater. The results also show that the enhancement of regeneration brought by ultrasound is not a simple sum of their respective contribution. © 2014 American Institute of Chemical Engineers AIChE J, 60: 1843–1853, 2014     

Linear track estimation by double pulse sources for moving target

The double pulse sources method (DPS) is used to estimate the track of the high speed underwater vehicle. The linear track parameters and the speed at each time will be achieved based on the direction estimated by two CW pulse reference sources, the emitted and recorded time of each pulse front, and the dual-source distance which is known in advance. Simulation results show that the DPS method improves the performance of the traditional source estimation reference method. In the reservoir simulation experiments, the deviation of direction estimation is less than 1％ and the deviation of distance estimation is controlled to within 9％ by the DPS algorithm. The DPS is suitable for underwater vehicle track estimation.