公开(公告)号：US08401852B2

公开(公告)日：2013-03-19

申请号：US12626943

申请日：2009-11-30

申请人： Geoffrey Gerson Zweig ,  Patrick An-Phu Nguyen ,  James Garnet Droppo, III ,  Alejandro Acero

发明人： Geoffrey Gerson Zweig ,  Patrick An-Phu Nguyen ,  James Garnet Droppo, III ,  Alejandro Acero

IPC分类号： G10L15/04

CPC分类号： G10L15/10 ,  G10L15/02

摘要： A computer-implemented speech recognition system described herein includes a receiver component that receives a plurality of detected units of an audio signal, wherein the audio signal comprises a speech utterance of an individual. A selector component selects a subset of the plurality of detected units that correspond to a particular time-span. A generator component generates at least one feature with respect to the particular time-span, wherein the at least one feature is one of an existence feature, an expectation feature, or an edit distance feature. Additionally, a statistical speech recognition model outputs at least one word that corresponds to the particular time-span based at least in part upon the at least one feature generated by the feature generator component.

摘要翻译： 本文描述的计算机实现的语音识别系统包括接收组件，其接收多个检测到的音频信号的单元，其中该音频信号包括个人的讲话语音。 选择器部件选择对应于特定时间跨度的多个检测单元的子集。 发生器组件相对于特定时间跨度产生至少一个特征，其中所述至少一个特征是存在特征，期望特征或编辑距离特征之一。 另外，统计语音识别模型至少部分地基于由特征生成器组件生成的至少一个特征来输出对应于特定时间跨度的至少一个单词。

公开(公告)号：US20110131046A1

公开(公告)日：2011-06-02

申请号：US12626943

申请日：2009-11-30

申请人： Geoffrey Gerson Zweig ,  Patrick An-Phu Nguyen ,  James Garnet Droppo, III ,  Alejandro Acero

发明人： Geoffrey Gerson Zweig ,  Patrick An-Phu Nguyen ,  James Garnet Droppo, III ,  Alejandro Acero

IPC分类号： G10L15/04

CPC分类号： G10L15/10 ,  G10L15/02

摘要： A computer-implemented speech recognition system described herein includes a receiver component that receives a plurality of detected units of an audio signal, wherein the audio signal comprises a speech utterance of an individual. A selector component selects a subset of the plurality of detected units that correspond to a particular time-span. A generator component generates at least one feature with respect to the particular time-span, wherein the at least one feature is one of an existence feature, an expectation feature, or an edit distance feature. Additionally, a statistical speech recognition model outputs at least one word that corresponds to the particular time-span based at least in part upon the at least one feature generated by the feature generator component.

摘要翻译： 本文描述的计算机实现的语音识别系统包括接收组件，其接收多个检测到的音频信号的单元，其中该音频信号包括个人的讲话语音。 选择器部件选择对应于特定时间跨度的多个检测单元的子集。 发生器组件相对于特定时间跨度产生至少一个特征，其中所述至少一个特征是存在特征，期望特征或编辑距离特征之一。 另外，统计语音识别模型至少部分地基于由特征生成器组件生成的至少一个特征来输出对应于特定时间跨度的至少一个单词。