公开(公告)号：US08005674B2

公开(公告)日：2011-08-23

申请号：US11775251

申请日：2007-07-10

申请人： Eric W Janke ,  Bin Jia

发明人： Eric W Janke ,  Bin Jia

IPC分类号： G10L15/06 ,  G10L15/10 ,  G10L15/14

CPC分类号： G10L15/063 ,  G10L15/14

摘要： A recognition model set is generated. A technique is described to take advantage of the logarithm likelihood of real data for cross entropy to measure the mismatch between a training data and a training data derived model, and compare such type of mismatches between class dependent models and class independent model for evidence of model replacement. By using change of cross entropies in the decision of adding class independent Gaussian Mixture Models (GMMs), the good performance of class dependent models is largely retained, while decreasing the size and complexity of the model.

摘要翻译： 生成识别模型集。 描述了一种技术，以利用实际数据的对数似然度来测量训练数据和训练数据导出模型之间的不匹配，并比较类依赖模型和类独立模型之间的类型，以证明模型 替代。 通过在添加类独立高斯混合模型（GMM）的决策中使用交叉熵的变化，大大保持了类依赖模型的良好性能，同时减小了模型的大小和复杂性。

公开(公告)号：US20080126094A1

公开(公告)日：2008-05-29

申请号：US11775251

申请日：2007-07-10

申请人： Eric W. Janke ,  Bin Jia

发明人： Eric W. Janke ,  Bin Jia

IPC分类号： G10L15/28

CPC分类号： G10L15/063 ,  G10L15/14

摘要： A method, system, and computer program for generating a recognition model set. A technique is described to take advantage of the logarithm likelihood of real data for cross entropy to measure the mismatch between a training data and a training data derived model, and compare such type of mismatches between class dependent models and class independent model for evidence of model replacement. By using change of cross entropies in the decision of adding class independent Gaussian Mixture Models (GMMs), the good performance of class dependent models is largely retained, while decreasing the size and complexity of the model.

摘要翻译： 一种用于生成识别模型集的方法，系统和计算机程序。 描述了一种技术，以利用实际数据的对数似然度来测量训练数据和训练数据导出模型之间的不匹配，并比较类依赖模型和类独立模型之间的类型，以证明模型 替代。 通过在添加类独立高斯混合模型（GMM）的决策中使用交叉熵的变化，大大保持了类依赖模型的良好性能，同时减小了模型的大小和复杂性。

公开(公告)号：US06999925B2

公开(公告)日：2006-02-14

申请号：US10007990

申请日：2001-11-13

申请人： Volker Fischer ,  Siegfried Kunzmann ,  Eric-W. Janke ,  A. Jon Tyrrell

发明人： Volker Fischer ,  Siegfried Kunzmann ,  Eric-W. Janke ,  A. Jon Tyrrell

IPC分类号： G10L15/06 ,  G10L15/08 ,  G10L15/27 ,  G10L15/18

CPC分类号： G10L15/07

摘要： The present invention provides a computerized method and apparatus for automatically generating from a first speech recognizer a second speech recognizer which can be adapted to a specific domain. The first speech recognizer can include a first acoustic model with a first decision network and corresponding first phonetic contexts. The first acoustic model can be used as a starting point for the adaptation process. A second acoustic model with a second decision network and corresponding second phonetic contexts for the second speech recognizer can be generated by re-estimating the first decision network and the corresponding first phonetic contexts based on domain-specific training data.