公开(公告)号：US07043430B1

公开(公告)日：2006-05-09

申请号：US10130490

申请日：2000-11-22

申请人： Grace Chung ,  Hong Chung Leung ,  Suk Hing Wong

发明人： Grace Chung ,  Hong Chung Leung ,  Suk Hing Wong

IPC分类号： G10L15/00

CPC分类号： G10L15/02 ,  G10L25/15 ,  G10L2015/027

摘要： A system and method for speaker independent speech recognition is provided that integrates spectral and tonal analysis in a sequential architecture. The system analyzes the spectral content of a spoken syllable, or group of syllables, (18) and generates a spectral score for each of a plurality of predicted syllables (46, 22). Time alignment information (36) for the predicted syllable(s) is then sequentially passed to a tonal modeling block (14) which performs an iterative fundamental frequency contour estimation for the spoken syllable(s). The tones of adjacent syllables, as well as the rate of change of the tonal information, is then used to generate a tonal score for each of the plurality of predicted syllables. The tonal score (34) is then arithmetically combined with (40) the spectral score (32) in order to generate an output prediction.

摘要翻译： 提供了一种用于讲话者独立语音识别的系统和方法，其将顺序架构中的频谱和色调分析整合。 该系统分析一个口语音节或一组音节（18）的频谱内容，并为多个预测音节（46,22）中的每一个生成频谱分数。 然后，将预测音节的时间对准信息（36）顺序地传递到音调建模块（14），该音调建模块（14）对所述音节进行迭代基频估计。 然后使用相邻音节的音调以及音调信息的变化率来为多个预测音节中的每一个生成音调分数。 然后将色调得分（34）与（40）光谱得分（32）算术结合，以便产生输出预测。

公开(公告)号：US08015013B2

公开(公告)日：2011-09-06

申请号：US11637357

申请日：2006-12-11

申请人： Chi Fai Chiu ,  Hing Fai Louis Chong ,  Ching Wai Jimmy Lee ,  Hong Chung Leung ,  Chi Wai Dennis Wee ,  Wei To William Wang

发明人： Chi Fai Chiu ,  Hing Fai Louis Chong ,  Ching Wai Jimmy Lee ,  Hong Chung Leung ,  Chi Wai Dennis Wee ,  Wei To William Wang

IPC分类号： G10L11/00 ,  G06F17/20 ,  G06F17/30

CPC分类号： G11B27/105 ,  G06F17/275 ,  G06F17/30026 ,  G10L15/18 ,  G10L15/26

摘要： There is provided a method for accessing at least one digital file from a collection comprising more than one digital file in an electronic device, including: generating one index comprising of information entries obtained from each of the more than one digital file in the collection, with each digital file in the collection information being linked to at least one information entry; receiving a speaker independent speech input in at least one language during a speech reception mode; determining a language of the speech input; and setting the speech reception mode to the language of the speech input; comparing the speech input received during the speech reception mode with the entries in the index. The file may advantageously be accessed when the speech input coincides with at least one of the information entries in the index. The digital files may be stored in the electronic device, any device functionally connected to the electronic device or a combination of the aforementioned. The at least one digital file may be received from a source selected from: a memory device, a wired computer network or a wireless computer network. An apparatus that is able to carry out the aforementioned method is also disclosed.

摘要翻译： 提供了一种用于从包括电子设备中的多于一个数字文件的集合访问至少一个数字文件的方法，包括：生成包括从集合中的多于一个数字文件中的每一个获得的信息条目的一个索引， 收集信息中的每个数字文件被链接到至少一个信息条目; 在语音接收模式期间以至少一种语言接收讲话者独立语音输入; 确定语音输入的语言; 并将语音接收模式设置为语音输入的语言; 将在语音接收模式期间接收到的语音输入与索引中的条目进行比较。 当语音输入与索引中的信息条目中的至少一个一致时，可以有利地访问该文件。 数字文件可以存储在电子设备中，功能上连接到电子设备的任何设备或者上述的组合。 可以从选自以下的源接收至少一个数字文件：存储设备，有线计算机网络或无线计算机网络。 还公开了能够执行上述方法的装置。