专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明申请

WO00013171A1 SPEAKER NORMALIZATION APPARATUS 审中-公开
标题翻译：扬声器正规化设备
公开(公告)号：WO00013171A1
公开(公告)日：2000-03-09
申请号：PCT/US1998/018013
申请日：1998-08-31
IPC分类号： G10L15/06 , G10L5/06 , G10L3/02
CPC分类号： G10L15/065
摘要： A signal conditioning apparatus and method to condition a human voice signal for input to a time-domain voice recognition system. The signal conditioner (10) normalizes sampled human voice inputs (12) so that all inputs result in an output with substantially the same pitch and frequency bandwidth characteristics. The signal conditioner (10) includes a pitch altering circuit (14) that alters the pitch of the input human voice signal (12) and a frequency band limiting circuit (16) that limits the range of spectral information contained in a signal. The pitch altering circuit (14) converts the input into a digital signal at a first conversion rate and converts the digital signal back to an analog signal at a second rate which is unequal to the first rate. The frequency band limit circuit (16) is a plurality of filters at different points in the frequency spectrum.
摘要翻译：一种信号调节装置和方法，用于调节人类语音信号以输入到时域语音识别系统。信号调节器（10）对采样的人类语音输入（12）进行归一化，使得所有输入导致具有基本上相同的音调和频率带宽特性的输出。信号调节器（10）包括改变输入人声信号（12）的音调的音调改变电路（14）和限制信号中包含的频谱信息的范围的频带限制电路（16）。音调改变电路（14）以第一转换速率将输入转换为数字信号，并以不同于第一速率的第二速率将数字信号转换回模拟信号。频带限制电路（16）是频谱的不同点的多个滤波器。

2. 发明申请

WO99059135A2 ARRANGEMENT AND METHOD FOR COMPUTER RECOGNITION OF A PREDEFINED VOCABULARY IN SPOKEN LANGUAGE 审中-公开
标题翻译：安排和方法，计算机口语给定的词汇检测
公开(公告)号：WO99059135A2
公开(公告)日：1999-11-18
申请号：PCT/DE1999/001323
申请日：1999-05-03
IPC分类号： G10L15/065 , G10L15/14 , G10L5/06 , G10L3/00
CPC分类号： G10L15/065 , G10L15/144
摘要： In speech recognition, phonemes of a language are modelled, whereby each state of the hidden Markov model is described by means of a probability density function. The probability density function is divided into a first and a second probability function in order to recognize modifications of vocabulary. This makes it possible to compensate for changes in the spoken habits of a speaker, enables a new word to be added to the vocabulary of the speech recognition system and guarantees that the new word can be recognized and distinguished from the words that are already present in the speech recognition system with a sufficient amount of quality.
摘要翻译：在一种语言的语音识别音素被隐马尔可夫模型模型化，由概率密度函数描述的隐马尔可夫模型的各状态。在词汇表中的变化的语音识别的概率密度函数被分成第一和第二概率密度函数。这使得以补偿扬声器的语言习惯的改变或添加新词来演讲的词汇，并确保这个新词是尊贵与演讲词已经存在的足够的质量，从而认可。

3. 发明申请

WO99056275A1 RELIABLE TEXT CONVERSION OF VOICE IN A RADIO COMMUNICATION SYSTEM AND METHOD 审中-公开
标题翻译：无线电通信系统和方法中语音的可靠文本转换
公开(公告)号：WO99056275A1
公开(公告)日：1999-11-04
申请号：PCT/US1999/006600
申请日：1999-03-25
IPC分类号： G10L15/00 , G10L15/26 , H04M1/64 , G10L5/06
CPC分类号： H04M3/533 , H04M2201/40 , H04M2201/60
摘要： A radio communication system includes a voice recognition system (218), a transmitter (202), and a processing system (210). The transmitter (202) is used for transmitting messages to a plurality of SCRs (selective call radios) (122). The processing system (210) is adapted to cause the voice recognition system (218) to convert a voice signal representative of a voice message originated by a caller to a text message (401, 417), wherein the text message is intended for a SCR (122), and to cause the transmitter (202) to transmit the text message (401, 417) to the SCR (122). An embodiment of the voice recognition system (218) may also generate a likelihood of success of flawless conversion (418), and the processing system will transmit the text message or prompt a human operator (424) to generate a corrected text message based on an accuracy threshold (422).
摘要翻译：无线电通信系统包括语音识别系统（218），发射机（202）和处理系统（210）。发射机（202）用于向多个SCR（选择性呼叫无线电）发送消息（122）。处理系统（210）适于使语音识别系统（218）将表示由呼叫者发起的语音消息的语音信号转换为文本消息（401,417），其中文本消息旨在用于SCR （122），并且使所述发送器（202）将所述文本消息（401,417）发送到所述SCR（122）。语音识别系统（218）的实施例还可以产生成功完成无缺陷转换的可能性（418），并且处理系统将发送文本消息或提示人类操作者（424）以基于精度阈值（422）。

4. 发明申请

WO99053478A1 DYNAMICALLY CONFIGURABLE ACOUSTIC MODEL FOR SPEECH RECOGNITION SYSTEMS 审中-公开
标题翻译：用于语音识别系统的动态配置声学模型
公开(公告)号：WO99053478A1
公开(公告)日：1999-10-21
申请号：PCT/US1999/006837
申请日：1999-03-29
IPC分类号： G10L15/02 , G10L15/06 , G10L15/14 , G10L15/26 , G10L5/06
CPC分类号： G10L15/26 , G10L15/144 , G10L2015/0631
摘要： The present invention includes a system for recognizing speech based on an input data stream. The system includes an acoustic model which has a model size. The model is adjustable to a desired size based on characteristics of a computer system on which the recognition system is run. The adjustment is obtained by clustering parameters of a senone tree.
摘要翻译：本发明包括一种基于输入数据流识别语音的系统。该系统包括具有模型尺寸的声学模型。基于运行识别系统的计算机系统的特性，该模型可以调整到所需的大小。调整是通过聚苯乙烯树的参数获得的。

5. 发明申请

WO9931654A3 METHOD OF DETERMINING MODEL-SPECIFIC FACTORS FOR PATTERN RECOGNITION, IN PARTICULAR FOR SPEECH PATTERNS 审中-公开
标题翻译：确定特定于模型的因素以进行模式识别，特别是语音模式的方法
公开(公告)号：WO9931654A3
公开(公告)日：1999-08-26
申请号：PCT/IB9801990
申请日：1998-12-11
申请人： KONINKL PHILIPS ELECTRONICS NV , PHILIPS SVENSKA AB , PHILIPS PATENTVERWALTUNG , BEYERLEIN PETER
发明人： BEYERLEIN PETER
IPC分类号： G10L15/10 , G06F17/17 , G10L15/14 , G10L15/183 , G10L5/06
CPC分类号： G10L15/183 , G06F17/17 , G10L15/142
摘要： A method for recognizing a pattern that comprises a set of physical stimuli, said method comprising the steps of: providing a set of training observations and through applying a plurality of association models ascertaining various measuring values pj(k ¦ x), j=1... M, that each pertain to assigning a particular training observation to one or more associated pattern classes; setting up a log/linear association distribution by combining all association models of the plurality according to respective weight factors, and joining thereto a normalization quantity to produce a compound association distribution; optimizing said weight factors for thereby minimizing a detected error rate of the actual assigning to said compound distribution; recognizing target observations representing a target pattern with the help of said compound distribution.
摘要翻译：一种用于识别包括一组物理刺激的图案的方法，所述方法包括以下步骤：提供一组训练观察并通过应用确定各种测量值pj（k | x），j = 1的多个关联模型。 ......每个都涉及为一个或多个相关模式类别分配特定的训练观察值; 通过根据各自的权重因子组合多个所述多个关联模型中的所有关联模型并且与其结合归一化量来建立对数/线性关联分布以产生复合关联分布; 优化所述权重因子，从而最小化实际分配给所述化合物分布的检测到的错误率; 在所述化合物分布的帮助下识别代表目标模式的目标观察值。

6. 发明申请

WO99030314A1 METHOD FOR VOICE RECOGNITION USING A GRAMMAR 审中-公开
标题翻译：方法的语音识别用某种语法
公开(公告)号：WO99030314A1
公开(公告)日：1999-06-17
申请号：PCT/DE1998/003536
申请日：1998-12-02
IPC分类号： G10L15/187 , G10L15/193 , G10L15/197 , G10L5/06
CPC分类号： G10L15/193 , G10L15/187 , G10L15/197
摘要： The invention relates to a method for voice recognition, wherein a bigram method with integrated unequivocal syntax restriction is combined with an N-gram voice model with statistical word sequence evaluation in such a way that alternative recognition methods can be used in different segments of a word sequence.
摘要翻译：对于用于识别语音的方法，具有集成的独特语法约束一个两字组方法，提出在该替代方案中的识别方法被施加到字序列的不同部分的方式与统计单词序列评价用的N-gram语言模型结合。

7. 发明申请

WO9926233A2 METHOD AND SYSTEM ARRANGED FOR SELECTIVE HARDWARE SHARING IN A SPEECH-BASED INTERCOMMUNICATION SYSTEM WITH SPEECH PROCESSING ON PLURAL LEVELS OF RELATIVE COMPLEXITY 审中-公开
标题翻译：在基于语音的对讲系统中选择性硬件共享的方法和系统与相对复杂度的平均水平进行语音处理
公开(公告)号：WO9926233A2
公开(公告)日：1999-05-27
申请号：PCT/IB9801651
申请日：1998-10-19
申请人： KONINKL PHILIPS ELECTRONICS NV , PHILIPS SVENSKA AB
发明人： JOOST MICHAEL
IPC分类号： G10L15/28 , G10L15/22 , G10L15/26 , G10L5/06
CPC分类号： G10L15/26 , G10L15/22
摘要： In a multistation intercommunication system human speech is processed on at least two respective levels of generic complexity. The speech is received in one or more origin stations in parallel and a necessity is detected to understand the speech in an associated application environment. Intercommunication is controlled in a distributed manner, by detecting temporal speech items to be recognized and dynamically assigning speech items amongst one or more of a plurality of distributed speech recognizing facilities to eventually generate recognized items. Further intercommunication is controlled to understand recognized items in a further context of the application in question through assigning the recognized items amongst one or more of a plurality of speech understanding facilities to generate speech items that have been understood. Assigning is effected in a distributed manner as based on a combination of contingency and statistics.
摘要翻译：在多通道互通系统中，人类语音在通用复杂度的至少两个相应级别上被处理。语音在并行的一个或多个起始站中被接收，并且检测必要性以理解相关应用环境中的语音。通过检测待识别的时间语音项目并在多个分布式语音识别设施中的一个或多个中动态地分配语音项目以最终生成识别的项目，来以分布式方式控制互通。通过在多个语音理解设施中的一个或多个语音理解设施中分配识别的项目以产生已经被理解的语音项目，进一步相互通信被控制以在所述应用的进一步上下文中理解所识别的项目。分配是按照偶然性和统计学的组合以分布式方式实现的。

8. 发明申请

WO99022364A1 SYSTEM AND METHOD FOR AUTOMATICALLY CLASSIFYING THE AFFECTIVE CONTENT OF SPEECH 审中-公开
标题翻译：用于自动分类语音的有效内容的系统和方法
公开(公告)号：WO99022364A1
公开(公告)日：1999-05-06
申请号：PCT/US1998/022836
申请日：1998-10-29
IPC分类号： G10L15/18 , G10L15/22 , G10L17/00 , G10L5/06
CPC分类号： G10L17/26 , G10L15/1807 , G10L2015/227
摘要： The classification (18) of the emotional content of speech (10) employs (16) acoustic measures (14) as classification input: pitch, spectral envelope shape (25) as either mel-frequency cepstral coefficients (MFCC) or linear prediction coefficients (LPC), variances in pitch and envelope, pitch magnitude, and energy. Changes in the spectral shape of the speech signal may distinguish long, smoothly varying sounds from quickly changing sound, which may indicate (18) the emotional state of the speaker. Different variations of pitch and spectral shape coefficients can be measured and analyzed to assist in the classification of individual utterances. Analysis can be performed on an entire utterance (28d), multiple segments of an utterance (10), or first, middle, and last thirds (28a, b, c) of an utterance.
摘要翻译：语音情感内容（10）的分类（18）采用（16）声学测量（14）作为分类输入：音调，频谱包络形状（25）作为梅尔频率倒谱系数（MFCC）或线性预测系数 LPC），音调和包络的方差，音调幅度和能量。语音信号的频谱形状的变化可以区分长，平滑变化的声音与快速变化的声音，这可能表明（18）说话者的情感状态。可以测量和分析音调和频谱形状系数的不同变化，以帮助分类各个话语。分析可以在话语的整个话语（28d），话语的多个部分（10）或第一，中，最后三分之二（28a，b，c）中进行。

9. 发明申请

WO99014743A1 SYSTEM AND METHOD USING NATURAL LANGUAGE UNDERSTANDING FOR SPEECH CONTROL APPLICATION 审中-公开
标题翻译：使用自然语言理解语音控制应用的系统和方法
公开(公告)号：WO99014743A1
公开(公告)日：1999-03-25
申请号：PCT/US1998/019433
申请日：1998-09-17
IPC分类号： G06F3/16 , G10L15/18 , G10L15/26 , G10L15/28 , G10L5/06
CPC分类号： G10L15/26 , G10L15/1815
摘要： The present invention is a computer apparatus and method for adding speech interpreting capabilities to an interactive voice response system. An annotated corpus is used to list valid utterances within a grammar along with token data for each valid utterance representing the meaning behind the valid utterance. When valid utterances are detected, the interactive voice response system requests that a search is made through the annotated corpus to find the token identified with the valid utterance. This token is returned to the interactive voice response system. If the valid utterance included a variable, additional processing is performed to interpret the variable and return additional data representing the variable.
摘要翻译：本发明是一种用于将语音解释能力添加到交互式语音响应系统的计算机装置和方法。注释语料库用于列出语法中的有效话语以及表示有效话语背后意义的每个有效语句的令牌数据。当检测到有效的话语时，交互式语音响应系统请求通过注释的语料库进行搜索以找到用有效话语标识的令牌。该令牌返回到交互式语音应答系统。如果有效的话语包括变量，则执行附加处理来解释变量并返回表示变量的附加数据。

10. 发明申请

WO99013456A1 SPEECH REFERENCE ENROLLMENT METHOD 审中-公开
标题翻译：语音参考引用方法
公开(公告)号：WO99013456A1
公开(公告)日：1999-03-18
申请号：PCT/US1998/017095
申请日：1998-08-17
IPC分类号： G10L15/06 , G10L17/00 , H04M3/38 , H04M3/493 , G10L5/06
CPC分类号： H04M3/382 , G10L15/07 , G10L17/04 , G10L2015/0631 , G10L2015/0636 , G10L2015/0638 , H04M3/493 , H04M2201/40
摘要： A speech reference enrollment method involves the following steps: (a) requesting a user speak a vocabulary word; (b) detecting a first utterance (354); (c) requesting the user speak the vocabulary word; (d) detecting a second utterance (358); (e) determining a first similarity between the first utterance and the second utterance (362); (f) when the first similarity is less than a predetermined similarity, requesting the user speak the vocabulary word; (g) detecting a third utterance (366); (h) determining a second similarity between the first utterance and the third utterance (370); and (i) when the second similarity is greater than or equal to the predetermined similarity, creating a reference (364).
摘要翻译：语音参考注册方法包括以下步骤：（a）请求用户说出词汇单词; （b）检测第一话语（354）; （c）请求用户说出词汇; （d）检测第二话语（358）; （e）确定第一话语和第二发音之间的第一相似度（362）; （f）当第一相似度小于预定相似度时，请求用户说出词汇单词; （g）检测第三个发音（366）; （h）确定所述第一话语和所述第三语音之间的第二相似度（370）; 和（i）当第二相似度大于或等于预定相似度时，创建参考（364）。

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式