面向口语统计语言模型建模的自动语料生成算法

在资源相对匮乏的自动语音识别(Automatic speech recognition, ASR)领域, 如面向电话交谈的语音识别系统中, 统计语言模型(Language model, LM)存在着严重的数据稀疏问题. 本文提出了一种基于等概率事件的采样语料生成算法, 自动生成领域相关的语料, 用来强化统计语言模型建模. 实验结果表明, 加入本算法生成的采样语料可以缓解语言模型的稀疏性, 从而提升整个语音识别系统的性能. 在开发集上语言模型的困惑度相对降低7.5%, 字错误率(Character error rate, CER)绝对降低0.2个点; 在测试集上语言模型的困惑度相对降低6%, 字错误率绝对降低0.4点.     

存储过程在ASP．NET数据库系统开发中的应用

基于ASP．NET的数据库系统开发应用越来越广泛,其中对数据库的操作是一个很重要的内容,而通过调用存储过程实现对数据库的操作是一个很实用的方法。文章将结合企业客户管理系统的开发来介绍存储过程在数据库项目开发中的具体应用。     

不完全匹配的语音和文本语句级对齐

语音文本自动对齐技术广泛应用于语音识别与合成、内容制作等领域,其主要目的是将语音和相应的参考文本在语句、单词、音素等级别的单元进行对齐,并获得语音与参考文本之间的时间对位信息.最新的先进对齐方法大多基于语音识别,一方面,准确率受限于语音识别效果,识别字错误率高时文语对齐精度明显下降,识别字错误率对对齐精度影响较大;另一方面,这种对齐方法不能有效处理不完全匹配的长篇幅语音和文本的对齐.该文提出一种基于锚点和韵律信息的文语对齐方法,通过基于边界锚点加权的片段标注将语料划分为对齐段和未对齐段,针对未对齐段使用双门限端点检测方法提取韵律信息,并检测语句边界,降低了基于语音识别的对齐方法对语音识别效果的依赖程度.实验结果表明,与目前先进的基于语音识别的文语对齐方法比较,即使在识别字错误率为0.52时,该文所提方法的对齐准确率仍能提升45%以上;在音频文本不匹配程度为0.5时,该文所提方法能提高3%.     

Harnessing graphics processors for the fast computation of acoustic likelihoods in speech recognition

In large vocabulary continuous speech recognition (LVCSR) the acoustic model computations often account for the largest processing overhead. Our weighted finite state transducer (WFST) based decoding engine can utilize a commodity graphics processing unit (GPU) to perform the acoustic computations to move this burden off the main processor. In this paper we describe our new GPU scheme that can achieve a very substantial improvement in recognition speed whilst incurring no reduction in recognition accuracy. We evaluate the GPU technique on a large vocabulary spontaneous speech recognition task using a set of acoustic models with varying complexity and the results consistently show by using the GPU it is possible to reduce the recognition time with largest improvements occurring in systems with large numbers of Gaussians. For the systems which achieve the best accuracy we obtained between 2.5 and 3 times speed-ups. The faster decoding times translate to reductions in space, power and hardware costs by only requiring standard hardware that is already widely installed.     

新闻视频结构化浏览与标注系统

阐述一种新颖的新闻视频结构化浏览和标注系统。应用基于时空切片分析的新闻主播检测方法和基于颜色直方图的镜头分割方法实现新闻视频的结构化。通过自动语音识别技术和特定语义概念模型的建立实现了对主播场景的文本信息标注和对新闻故事镜头的语义概念标注。该系统有利于用户根据个人爱好进行新闻视频的浏览和编辑,有效实现新闻视频的索引和浏览。     

AS-R移动机器人的动态避障与路径规划研究

针对移动机器人的动态避障和路径规划问题,以AS-R移动机器人为平台,设计了一种基于行为分析的动态避障策略。根据避障问题将移动机器人整个运行过程中的行为划分成趋向目标行为、避障行为、沿墙走行为及紧急避障4种行为,有效实现了机器人的动态避障,并解决了两种避障难题:左右摆动问题和凹型障碍物问题。利用多传感器结合检测方法,通过红外传感器减少盲区和镜面反射带来的误差,通过间隔采样或分组采样技术避免多路串扰问题;对均值滤波与中值滤波进行实验对比后,提出一种递推型中值滤波方法,从而提高了数据在空间和时间上的连续性,有效地减少了超声波随机串扰信号及其它干扰信号,进而提高了探测模块的准确度。最后设计了几种复杂环境下机器人的动态避障和路径规划,并验证了所提方法的有效性。     

信息系统中基于角色的访问控制

在一个企业级信息系统中,用户访问信息系统中资源的管理是一个复杂问题,直接对用户授权的管理模式不能适应企业的发展。文中讨论了基于角色的访问控制管理方式和ORACLE数据库管理系统的安全性策略,分析了信息系统中访问控制管理的需求,提出了对信息系统应用功能访问和数据库访问的分开控制管理,通过应用系统角色建立数据库用户和应用系统功能之间的关联,从而构成信息系统的访问控制管理策略。给出了基于角色的访问控制实现,并应用于设备信息管理系统中,实现了用户访问控制动态管理。     

Fiber reinforced mortar affected by alkali-silica reaction: A study by synchrotron microtomography

Alkali-Silica reaction (ASR) is a physicochemical process that can deteriorate concrete and is a recurring engineering problem. In this study three different cylindrical samples affected by ASR were prepared: a plain mortar and two composite mortars containing fibers (polypropylene and a polymer hybrid), which were analyzed at the microtomography (μCT) beamline 8.3.2 at the Advanced Light Source (ALS). In general, three different features were observed during the 136 day observation period: (1) aggregate dissolution, (2) crack propagation from inside the aggregate, through the cement matrix, and at the ITZ, and (3) the alkali-silica gel filling cracks and voids. In addition, accelerated mortar bar tests were utilized to observe ASR's expansive effect in the plain and composite mortars, and the fibers' ability to restrain expansion due to ASR.     

Effect of microstructure on the electrochemical performance of Ni-ScSZ anodes

The effects of NiO powder morphology and sintering temperature on the microstructure and the electrochemical performance of Nickel-scandia-stabilized zirconia (Ni-ScSZ) cermet anodes for solid oxide fuel cells (SOFCs) were investigated. The particle size and agglomeration of the starting powders were found to affect both the microstructure and electrochemical performance of the Ni-ScSZ cermet anodes. The lowest polarization resistance, 0.690 Ω cm² at 700 °C, was measured for the Ni-ScSZ anode prepared with fine NiO powder (~0.5 µm grain size). This was attributed to the increase in the number of reaction sites afforded by the small grains and well-dispersed Ni and ScSZ phases. The effect of the anode sintering temperature was also found to affect the anode microstructure, adhesion with the electrolyte, and consequently anode polarization resistance. The lowest polarization resistance was observed for the anode sintered at 1400 °C and this was 3–5 times lower than the corresponding values for anodes sintered at lower temperatures.     

粉煤灰抑制集料碱硅酸反应的有效性评估

可靠地评价粉煤灰抑制ASR的有效性是工程中使用粉煤灰预防ABR的关键所在。可从粉煤灰抑制ASR的能力、抑制ASR膨胀的程度以及抑制ASR膨胀的工程安全性等3个方面对掺粉煤灰抑制ASR的有效性进行评估。评估结果表明，掺20％的粉煤灰可以有效抑制活性集料ASR的膨胀。