公开(公告)号：US20080181489A1

公开(公告)日：2008-07-31

申请号：US11700664

申请日：2007-01-31

申请人： Qiang Fu ,  Xiaodong He ,  Li Deng

发明人： Qiang Fu ,  Xiaodong He ,  Li Deng

IPC分类号： G06K9/62

CPC分类号： G06K9/6217 ,  G10L15/142

摘要： Pattern model parameters are updated using update equations based on competing patterns that are identical to a reference pattern except for one segment at a time that is replaced with a competing segment. This allows pattern recognition parameters to be tuned one segment at a time, rather than have to try to model distinguishing features of the correct pattern model as a whole, according to an illustrative embodiment. A reference pattern and competing patterns are divided into pattern segments. A set of training patterns are generated by replacing one of the pattern segments in the reference pattern with a corresponding competing pattern segment. For each of the training patterns, a pattern recognition model is applied to evaluate a relative degree of correspondence of the reference pattern with the pattern signal compared to a degree of correspondence of the training patterns with the pattern signal.

摘要翻译： 基于与参考模式相同的竞争模式的更新方程来更新模式模型参数，除了一次被竞争的段替换的一个段。 这允许模式识别参数一次调整一个段，而不是根据说明性实施例而不必为整体模拟正确模式模型的区分特征。 参考模式和竞争模式分为模式段。 通过将参考图案中的一个图案片段替换为相应的竞争图案片段来生成一组训练图案。 对于每个训练模式，应用模式识别模型来评估参考模式与模式信号的相对程度，与训练模式与模式信号的对应程度相比较。

公开(公告)号：US07873209B2

公开(公告)日：2011-01-18

申请号：US11700664

申请日：2007-01-31

申请人： Li Deng ,  Xiaodong He ,  Qiang Fu

发明人： Li Deng ,  Xiaodong He ,  Qiang Fu

IPC分类号： G06K9/00 ,  G10L15/00

CPC分类号： G06K9/6217 ,  G10L15/142

摘要： Pattern model parameters are updated using update equations based on competing patterns that are identical to a reference pattern except for one segment at a time that is replaced with a competing segment. This allows pattern recognition parameters to be tuned one segment at a time, rather than have to try to model distinguishing features of the correct pattern model as a whole, according to an illustrative embodiment. A reference pattern and competing patterns are divided into pattern segments. A set of training patterns are generated by replacing one of the pattern segments in the reference pattern with a corresponding competing pattern segment. For each of the training patterns, a pattern recognition model is applied to evaluate a relative degree of correspondence of the reference pattern with the pattern signal compared to a degree of correspondence of the training patterns with the pattern signal.

摘要翻译： 基于与参考模式相同的竞争模式的更新方程来更新模式模型参数，除了一次被竞争的段替换的一个段。 这允许模式识别参数一次调整一个段，而不是根据说明性实施例而不必为整体模拟正确模式模型的区分特征。 参考模式和竞争模式分为模式段。 通过将参考图案中的一个图案片段替换为相应的竞争图案片段来生成一组训练图案。 对于每个训练模式，应用模式识别模型来评估参考模式与模式信号的相对程度，与训练模式与模式信号的对应程度相比较。

公开(公告)号：US08423364B2

公开(公告)日：2013-04-16

申请号：US11708440

申请日：2007-02-20

申请人： Dong Yu ,  Alejandro Acero ,  Li Deng ,  Xiaodong He

发明人： Dong Yu ,  Alejandro Acero ,  Li Deng ,  Xiaodong He

IPC分类号： G10L15/14 ,  G10L15/00 ,  G10L15/06

CPC分类号： G10L15/063 ,  G10L2015/0631

摘要： A method and apparatus for training an acoustic model are disclosed. A training corpus is accessed and converted into an initial acoustic model. Scores are calculated for a correct class and competitive classes, respectively, for each token given the initial acoustic model. Also, a sample-adaptive window bandwidth is calculated for each training token. From the calculated scores and the sample-adaptive window bandwidth values, loss values are calculated based on a loss function. The loss function, which may be derived from a Bayesian risk minimization viewpoint, can include a margin value that moves a decision boundary such that token-to-boundary distances for correct tokens that are near the decision boundary are maximized. The margin can either be a fixed margin or can vary monotonically as a function of algorithm iterations. The acoustic model is updated based on the calculated loss values. This process can be repeated until an empirical convergence is met.

摘要翻译： 公开了一种用于训练声学模型的方法和装置。 训练语料库被访问并转换成初始声学模型。 对于给定初始声学模型的每个令牌，分数计算分别为正确的类和竞争类。 此外，针对每个训练令牌计算样本自适应窗口带宽。 从计算出的分数和采样自适应窗口带宽值，根据损失函数计算损失值。 可以从贝叶斯风险最小化观点导出的损失函数可以包括移动判定边界的边距值，使得靠近判定边界的正确令牌的令牌到边界的距离最大化。 边距可以是固定边距，也可以作为算法迭代的函数单调变化。 基于计算的损失值更新声学模型。 可以重复该过程，直到满足经验收敛。

公开(公告)号：US08301449B2

公开(公告)日：2012-10-30

申请号：US11581673

申请日：2006-10-16

申请人： Xiaodong He ,  Li Deng

发明人： Xiaodong He ,  Li Deng

IPC分类号： G10L15/00

CPC分类号： G10L15/063 ,  G10L15/144

摘要： Hidden Markov Model (HMM) parameters are updated using update equations based on growth transformation optimization of a minimum classification error objective function. Using the list of N-best competitor word sequences obtained by decoding the training data with the current-iteration HMM parameters, the current HMM parameters are updated iteratively. The updating procedure involves using weights for each competitor word sequence that can take any positive real value. The updating procedure is further extended to the case where a decoded lattice of competitors is used. In this case, updating the model parameters relies on determining the probability for a state at a time point based on the word that spans the time point instead of the entire word sequence. This word-bound span of time is shorter than the duration of the entire word sequence and thus reduces the computing time.

摘要翻译： 使用基于最小分类误差目标函数的生长变换优化的更新方程来更新隐马尔可夫模型（HMM）参数。 使用通过使用当前迭代HMM参数对训练数据进行解码而获得的N个最佳竞争者词序列表，迭代地更新当前HMM参数。 更新过程涉及使用可以获得任何正实值的每个竞争者词序列的权重。 更新过程进一步扩展到使用竞争者的解码格子的情况。 在这种情况下，更新模型参数依赖于基于跨越时间点而不是整个单词序列的单词来确定在时间点的状态的概率。 这个字边界的时间范围比整个单词序列的持续时间短，从而减少了计算时间。

公开(公告)号：US08407041B2

公开(公告)日：2013-03-26

申请号：US12957394

申请日：2010-12-01

申请人： Li Deng ,  Yaodong Zhang ,  Alejandro Acero ,  Xiaodong He

发明人： Li Deng ,  Yaodong Zhang ,  Alejandro Acero ,  Xiaodong He

IPC分类号： G06F17/28

CPC分类号： G06F17/2818 ,  G10L15/14 ,  G10L15/183

摘要： Architecture that provides the integration of automatic speech recognition (ASR) and machine translation (MT) components of a full speech translation system. The architecture is an integrative and discriminative approach that employs an end-to-end objective function (the conditional probability of the translated sentence (target) given the source language's acoustic signal, as well as the associated BLEU score in the translation, as a goal in the integrated system. This goal defines the theoretically correct variables to determine the speech translation system output using a Bayesian decision rule. These theoretically correct variables are modified in practical use due to known imperfections of the various models used in building the full speech translation system. The disclosed approach also employs automatic training of these variables using minimum classification error (MCE) criterion. The measurable BLEU scores are used to facilitate the implementation of the MCE training procedure in a step that defines the class-specific discriminant function.

摘要翻译： 提供完整语音翻译系统的自动语音识别（ASR）和机器翻译（MT）组件的集成的架构。 该架构是一种综合和歧视性的方法，采用端到端目标函数（给定源语言的声信号的翻译句子（目标）的条件概率）以及翻译中相关联的BLEU得分作为目标 这个目标定义了理论上正确的变量来确定使用贝叶斯判决规则的语音翻译系统输出，这些理论上正确的变量在实际应用中被修改，这是由于建立全语音翻译系统中使用的各种模型的已知缺陷 所公开的方法还采用最小分类误差（MCE）标准对这些变量进行自动训练，可测量的BLEU分数用于在定义特定类别判别函数的步骤中促进MCE训练过程的实现。

公开(公告)号：US20120143591A1

公开(公告)日：2012-06-07

申请号：US12957394

申请日：2010-12-01

申请人： Li Deng ,  Yaodong Zhang ,  Alejandro Acero ,  Xiaodong He

发明人： Li Deng ,  Yaodong Zhang ,  Alejandro Acero ,  Xiaodong He

IPC分类号： G06F17/28

CPC分类号： G06F17/2818 ,  G10L15/14 ,  G10L15/183

摘要： Architecture that provides the integration of automatic speech recognition (ASR) and machine translation (MT) components of a full speech translation system. The architecture is an integrative and discriminative approach that employs an end-to-end objective function (the conditional probability of the translated sentence (target) given the source language's acoustic signal, as well as the associated BLEU score in the translation, as a goal in the integrated system. This goal defines the theoretically correct variables to determine the speech translation system output using a Bayesian decision rule. These theoretically correct variables are modified in practical use due to known imperfections of the various models used in building the full speech translation system. The disclosed approach also employs automatic training of these variables using minimum classification error (MCE) criterion. The measurable BLEU scores are used to facilitate the implementation of the MCE training procedure in a step that defines the class-specific discriminant function.

摘要翻译： 提供完整语音翻译系统的自动语音识别（ASR）和机器翻译（MT）组件的集成的架构。 该架构是一种综合和歧视性的方法，采用端到端目标函数（给定源语言的声信号的翻译句子（目标）的条件概率）以及翻译中相关联的BLEU得分作为目标 这个目标定义了理论上正确的变量来确定使用贝叶斯判决规则的语音翻译系统输出，这些理论上正确的变量在实际应用中被修改，这是由于建立全语音翻译系统中使用的各种模型的已知缺陷 所公开的方法还采用最小分类误差（MCE）标准对这些变量进行自动训练，可测量的BLEU分数用于在定义特定类别判别函数的步骤中促进MCE训练过程的实现。

公开(公告)号：US07617103B2

公开(公告)日：2009-11-10

申请号：US11509980

申请日：2006-08-25

申请人： Xiaodong He ,  Alex Acero ,  Dong Yu ,  Li Deng

发明人： Xiaodong He ,  Alex Acero ,  Dong Yu ,  Li Deng

IPC分类号： G10L15/14

CPC分类号： G10L15/063 ,  G10L15/144

摘要： A method and apparatus for training an acoustic model are disclosed. A training corpus is accessed and converted into an initial acoustic model. Scores are calculated for a correct class and competitive classes, respectively, for each token given the acoustic model. From this score a misclassification measure is calculated and then a loss function is calculated from the misclassification measure. The loss function also includes a margin value that varies over each iteration in the training. Based on the calculated loss function the acoustic model is updated, where the loss function with the margin value is minimized. This process repeats until such time as an empirical convergence is met.

摘要翻译： 公开了一种用于训练声学模型的方法和装置。 训练语料库被访问并转换成初始声学模型。 对于给定声学模型的每个令牌，分数是针对正确的班级和竞赛班分别计算的。 从该分数计算错误分类度量，然后根据误分类度量计算损失函数。 损失函数还包括在训练中每次迭代变化的保证金值。 基于计算的损耗函数，声学模型被更新，其中具有边际值的损失函数被最小化。 该过程重复，直到满足经验收敛的时间为止。

公开(公告)号：US20080201139A1

公开(公告)日：2008-08-21

申请号：US11708440

申请日：2007-02-20

申请人： Dong Yu ,  Alejandro Acero ,  Li Deng ,  Xiaodong He

发明人： Dong Yu ,  Alejandro Acero ,  Li Deng ,  Xiaodong He

IPC分类号： G10L15/00

CPC分类号： G10L15/063 ,  G10L2015/0631

摘要： A method and apparatus for training an acoustic model are disclosed. A training corpus is accessed and converted into an initial acoustic model. Scores are calculated for a correct class and competitive classes, respectively, for each token given the initial acoustic model. Also, a sample-adaptive window bandwidth is calculated for each training token. From the calculated scores and the sample-adaptive window bandwidth values, loss values are calculated based on a loss function. The loss function, which may be derived from a Bayesian risk minimization viewpoint, can include a margin value that moves a decision boundary such that token-to-boundary distances for correct tokens that are near the decision boundary are maximized. The margin can either be a fixed margin or can vary monotonically as a function of algorithm iterations. The acoustic model is updated based on the calculated loss values. This process can be repeated until an empirical convergence is met.

摘要翻译： 公开了一种用于训练声学模型的方法和装置。 训练语料库被访问并转换成初始声学模型。 对于给定初始声学模型的每个令牌，分数计算分别为正确的类和竞争类。 此外，针对每个训练令牌计算样本自适应窗口带宽。 从计算出的分数和采样自适应窗口带宽值，根据损失函数计算损失值。 可以从贝叶斯风险最小化观点导出的损失函数可以包括移动判定边界的边距值，使得靠近判定边界的正确令牌的令牌到边界的距离最大化。 边距可以是固定边距，也可以作为算法迭代的函数单调变化。 基于计算的损失值更新声学模型。 可以重复该过程，直到满足经验收敛。

公开(公告)号：US20130110491A1

公开(公告)日：2013-05-02

申请号：US13283633

申请日：2011-10-28

申请人： Xiaodong He ,  Li Deng

发明人： Xiaodong He ,  Li Deng

IPC分类号： G06F17/28

CPC分类号： G06F17/2818 ,  G10L15/18

摘要： Architecture that formulates speech translation as a unified log-linear model with a plurality of feature functions, some of which are derived from generative models. The architecture employs discriminative training for the generative features based on an optimization technique referred to as growth transformation. A discriminative training objective function is formulated for speech translation as well as a growth transformation-based model training method that includes an iterative training formula. This architecture is used to design and perform the global end-to-end optimization of speech translation, which when compared with conventional methods for speech translation provides not only a learning method with faster convergence but also improves speech translation accuracy.

摘要翻译： 将语音翻译制定为具有多个特征函数的统一对数线性模型的架构，其中一些特征函数源自生成模型。 该架构采用基于称为增长转型的优化技术的生成特征的辨别性训练。 为语音翻译制定了歧视性的训练目标函数，以及包含迭代训练公式的基于生长变换的模型训练方法。 该架构用于设计和执行语音翻译的全局端到端优化，与传统的语音翻译方法相比，语音翻译不仅提供了一种具有更快融合的学习方法，而且提高了语音翻译的准确性。

公开(公告)号：US20080091424A1

公开(公告)日：2008-04-17

申请号：US11581673

申请日：2006-10-16

申请人： Xiaodong He ,  Li Deng

发明人： Xiaodong He ,  Li Deng

IPC分类号： G10L15/00

CPC分类号： G10L15/063 ,  G10L15/144

摘要： Hidden Markov Model (HMM) parameters are updated using update equations based on growth transformation optimization of a minimum classification error objective function. Using the list of N-best competitor word sequences obtained by decoding the training data with the current-iteration HMM parameters, the current HMM parameters are updated iteratively. The updating procedure involves using weights for each competitor word sequence that can take any positive real value. The updating procedure is further extended to the case where a decoded lattice of competitors is used. In this case, updating the model parameters relies on determining the probability for a state at a time point based on the word that spans the time point instead of the entire word sequence. This word-bound span of time is shorter than the duration of the entire word sequence and thus reduces the computing time.

摘要翻译： 使用基于最小分类误差目标函数的生长变换优化的更新方程来更新隐马尔可夫模型（HMM）参数。 使用通过使用当前迭代HMM参数对训练数据进行解码而获得的N个最佳竞争者词序列表，迭代地更新当前HMM参数。 更新过程涉及使用可以获得任何正实值的每个竞争者词序列的权重。 更新过程进一步扩展到使用竞争者的解码格子的情况。 在这种情况下，更新模型参数依赖于基于跨越时间点而不是整个单词序列的单词来确定在时间点的状态的概率。 这个字边界的时间范围比整个单词序列的持续时间短，从而减少了计算时间。