会员体验
专利管家(专利管理)
工作空间(专利管理)
风险监控(情报监控)
数据分析(专利分析)
侵权分析(诉讼无效)
联系我们
交流群
官方交流:
QQ群: 891211   
微信请扫码    >>>
现在联系顾问~
热词
    • 1. 发明申请
    • Speech model refinement with transcription error detection
    • 语音模型细化与转录错误检测
    • US20080270133A1
    • 2008-10-30
    • US11789132
    • 2007-04-24
    • Ye TianYifan GongFrank K. Soong
    • Ye TianYifan GongFrank K. Soong
    • G10L15/00
    • G10L15/065G10L15/01G10L15/08G10L2015/0631
    • Reliable transcription error-checking algorithm that uses a word confidence score and a word duration probability to detect transcription errors for improved results through the automatic detection of transcription errors in a corpus. The transcription error-checking algorithm is combined model training so as to use a current model to detect transcription errors, remove utterances which contain incorrect transcription (or manually fix the found errors), and retrain the model. This process can be repeated for several iterations to obtain an improved speech recognition model. The speech model is employed to achieve speech-transcription alignment to obtain a word boundary. Speech recognizer is then utilized to generate a word-lattice. Using the word boundary and word lattice, error detection is computed using a word confidence score and a word duration probability.
    • 可靠的转录错误检查算法,通过自动检测语料库中的转录错误,使用单词置信度分数和单词持续时间概率来检测转录错误以改善结果。 转录错误检查算法是组合模型训练,以便使用当前模型来检测转录错误,删除包含不正确转录(或手动修复发现的错误)的话语,并重新训练模型。 该过程可以重复几次迭代以获得改进的语音识别模型。 语音模型用于实现语音转录对齐以获得字边界。 然后利用语音识别器来产生一个单词格。 使用单词边界和单词格,使用单词置信分数和单词持续时间概率来计算错误检测。
    • 2. 发明授权
    • Speech model refinement with transcription error detection
    • 语音模型细化与转录错误检测
    • US07860716B2
    • 2010-12-28
    • US11789132
    • 2007-04-24
    • Ye TianYifan GongFrank K. Soong
    • Ye TianYifan GongFrank K. Soong
    • G10L15/10
    • G10L15/065G10L15/01G10L15/08G10L2015/0631
    • Reliable transcription error-checking algorithm that uses a word confidence score and a word duration probability to detect transcription errors for improved results through the automatic detection of transcription errors in a corpus. The transcription error-checking algorithm is combined model training so as to use a current model to detect transcription errors, remove utterances which contain incorrect transcription (or manually fix the found errors), and retrain the model. This process can be repeated for several iterations to obtain an improved speech recognition model. The speech model is employed to achieve speech-transcription alignment to obtain a word boundary. Speech recognizer is then utilized to generate a word-lattice. Using the word boundary and word lattice, error detection is computed using a word confidence score and a word duration probability.
    • 可靠的转录错误检查算法,通过自动检测语料库中的转录错误,使用单词置信度分数和单词持续时间概率来检测转录错误以改善结果。 转录错误检查算法是组合模型训练,以便使用当前模型来检测转录错误,删除包含不正确转录(或手动修复发现的错误)的话语,并重新训练模型。 该过程可以重复几次迭代以获得改进的语音识别模型。 语音模型用于实现语音转录对齐以获得字边界。 然后利用语音识别器来产生一个单词格。 使用单词边界和单词格,使用单词置信分数和单词持续时间概率来计算错误检测。
    • 3. 发明申请
    • MODEL DEVELOPMENT AUTHORING, GENERATION AND EXECUTION BASED ON DATA AND PROCESSOR DEPENDENCIES
    • 基于数据和处理器依赖的模型开发创作,生成和执行
    • US20090177471A1
    • 2009-07-09
    • US11971897
    • 2008-01-09
    • Yifan GongYe Tian
    • Yifan GongYe Tian
    • G10L15/00
    • G10L15/28G06F2209/485G06K9/6253G06Q10/0637
    • A recognition (e.g., speech, handwriting, etc.) model build process that is declarative and data-dependence-based. Process steps are defined in a declarative language as individual processors having input/output data relationships and data dependencies of predecessors and subsequent process steps. A compiler is utilized to generate the model building sequence. The compiler uses the input data and output data files of each model build processor to determine the sequence of model building and automatically orders the processing steps based on the declared input/output relationship (the user does not need to determine the order of execution). The compiler also automatically detects ill-defined processes, including cyclic definition and data being produced by more than one action. The user can add, change and/or modify a process by editing a declaration file, and rerunning the compiler, thereby a new process is automatically generated.
    • 基于声明和数据依赖的识别(例如,语音,手写等)模型构建过程。 处理步骤以声明性语言定义为具有输入/输出数据关系以及前辈和后续处理步骤的数据依赖性的各个处理器。 利用编译器生成模型构建序列。 编译器使用每个模型构建处理器的输入数据和输出数据文件来确定模型构建的顺序,并根据声明的输入/输出关系自动排序处理步骤(用户不需要确定执行顺序)。 编译器还自动检测不明确的进程,包括由多个动作产生的循环定义和数据。 用户可以通过编辑声明文件来添加,更改和/或修改进程,并重新运行编译器,从而自动生成新的进程。
    • 4. 发明授权
    • Model development authoring, generation and execution based on data and processor dependencies
    • 基于数据和处理器依赖的模型开发创作,生成和执行
    • US08086455B2
    • 2011-12-27
    • US11971897
    • 2008-01-09
    • Yifan GongYe Tian
    • Yifan GongYe Tian
    • G10L15/12
    • G10L15/28G06F2209/485G06K9/6253G06Q10/0637
    • A recognition (e.g., speech, handwriting, etc.) model build process that is declarative and data-dependence-based. Process steps are defined in a declarative language as individual processors having input/output data relationships and data dependencies of predecessors and subsequent process steps. A compiler is utilized to generate the model building sequence. The compiler uses the input data and output data files of each model build processor to determine the sequence of model building and automatically orders the processing steps based on the declared input/output relationship (the user does not need to determine the order of execution). The compiler also automatically detects ill-defined processes, including cyclic definition and data being produced by more than one action. The user can add, change and/or modify a process by editing a declaration file, and rerunning the compiler, thereby a new process is automatically generated.
    • 基于声明和数据依赖的识别(例如,语音,手写等)模型构建过程。 处理步骤以声明性语言定义为具有输入/输出数据关系以及前辈和后续处理步骤的数据依赖性的各个处理器。 利用编译器生成模型构建序列。 编译器使用每个模型构建处理器的输入数据和输出数据文件来确定模型构建的顺序,并根据声明的输入/输出关系自动排序处理步骤(用户不需要确定执行顺序)。 编译器还自动检测不明确的进程,包括由多个动作产生的循环定义和数据。 用户可以通过编辑声明文件来添加,更改和/或修改进程,并重新运行编译器,从而自动生成新的进程。
    • 9. 发明申请
    • INTERNET ADDRESS INFORMATION PROCESSING METHOD, APPARATUS, AND INTERENT SYSTEM
    • 互联网地址信息处理方法,装置和国际系统
    • US20120290700A1
    • 2012-11-15
    • US13554992
    • 2012-07-20
    • Xiaodong LiWei MaoYe TianWei WangTao ChenDi Ma
    • Xiaodong LiWei MaoYe TianWei WangTao ChenDi Ma
    • G06F15/16
    • H04L61/6059H04L61/10H04L61/103H04L61/2007H04L61/6068
    • Embodiments of the present invention provide an Internet address information processing method, apparatus and an Internet system. The method includes: receiving, by a first leaf node, a query request message containing an IP address to be queried sent by user equipment, and performing query or sending a query request to an intermediate layer node; obtaining, by the intermediate layer node, an IP address of a second leaf node according to a prefix of the IP address to be queried to send the query request message to the second leaf node; and finally, sending, by the second leaf node, address information about the IP address after receiving the query request message. Embodiments of the present invention further provide corresponding apparatus and system. The method, apparatus, and system provided by the present invention are capable of implementing real-time query for address information about IP addresses.
    • 本发明的实施例提供一种因特网地址信息处理方法,装置和因特网系统。 该方法包括:由第一叶节点接收包含由用户设备发送的要查询的IP地址的查询请求消息,并向中间层节点执行查询或发送查询请求; 根据要查询的IP地址的前缀,由中间层节点获取第二叶节点的IP地址,以向第二叶节点发送查询请求消息; 最后,在接收到查询请求消息之后,由第二叶节点发送关于IP地址的地址信息。 本发明的实施例还提供相应的装置和系统。 本发明提供的方法,装置和系统能够实现关于IP地址的地址信息的实时查询。
    • 10. 发明授权
    • Inlet throttle controlled liquid pump with cavitation damage avoidance feature
    • 进气节流控制液泵具有气蚀损伤避免功能
    • US08202064B2
    • 2012-06-19
    • US12951093
    • 2010-11-22
    • Ye TianDavid C. MackAlan R. Stockner
    • Ye TianDavid C. MackAlan R. Stockner
    • F04B49/22F02M57/02
    • F04B49/225F02M59/205F02M59/34F02M59/464F02M2200/04F04B11/0091Y10T137/7784
    • A liquid pump includes an electronically controlled throttle inlet valve to control pump output. With each reciprocation cycle, a plunger displaces a fixed volume of fluid. When less than this fixed volume is desired as the output from the pump, the electronically controlled throttle inlet valve throttles flow past a passive inlet check valve to reduce output. As a consequence, cavitation bubbles are generated during the intake stroke. Cavitation damage to surfaces that define the inlet port passage are avoided by a specifically shaped and sized cavitation flow adjuster extending from the valve member of the passive inlet check valve. By positioning the cavitation flow adjuster in the inlet port passage, a flow pattern is formed in a way to encourage cavitation bubble collapse away from surfaces that could result in unacceptable cavitation damage to the pump.
    • 液体泵包括一个电子控制节气门入口阀,用于控制泵的输出。 在每个往复循环中,柱塞移动固定体积的流体。 当需要小于该固定体积作为来自泵的输出时,电子控制节气门入口阀节流阀流过被动进口止回阀以减少输出。 因此,在进气冲程期间产生气蚀气泡。 通过从被动入口止回阀的阀构件延伸的特定形状和尺寸的气蚀流量调节器来避免限定入口端口通道的表面的气蚀损坏。 通过将空化流调节器定位在入口通道中,形成流动模式以鼓励气泡从表面塌陷,这可能导致对泵的不可接受的空化损伤。