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

WO2011046474A3 METHOD FOR IDENTIFYING A SPEAKER BASED ON RANDOM SPEECH PHONOGRAMS USING FORMANT EQUALIZATION 审中-公开
标题翻译：基于使用均匀化的随机语音信使识别扬声器的方法
公开(公告)号：WO2011046474A3
公开(公告)日：2011-06-16
申请号：PCT/RU2010000661
申请日：2010-11-03
申请人： OBSCHESTVO S OGRANICHENNOI OTVETSTVENNOST YU CENTR RECHEVYH TEHNOLOGIJ , KOVAL SERGEY LVOVICH
发明人： KOVAL SERGEY LVOVICH
IPC分类号： G10L17/00 , G10L17/02 , G10L17/06 , G10L17/20 , G10L25/15
CPC分类号： G10L17/14 , G10L17/02 , G10L17/06 , G10L17/20 , G10L25/15
摘要： The invention relates to a method for the reliable identification of a speaker based on long or short phonograms, phonograms recorded in different channels with high interference and distortion levels, and phonograms including random speech of narrators under various psycho-physiological conditions and speaking in different languages. The present method can be used in a wide range of applications, including in criminal investigations. The identification of a speaker based on speech phonograms is made by estimating the convergence between a first phonogram of the speaker and a second calibration phonogram. In order to carry out the estimation, the method comprises selecting on the first and second phonograms reference fragments of speech signals containing formant trajectories of at least three formants, comparing the reference fragments in which the values of at least two formant frequencies coincide, estimating the convergence of the compared reference fragments based on the coincidence of the values of the remaining formant frequencies, and determining the convergence of all the phonograms based on the global estimation of the convergence of all the compared reference fragments.
摘要翻译：本发明涉及一种用于可靠地识别基于长或短录音的扬声器的方法，记录在具有高干扰和失真水平的不同频道中的录音，以及包括在各种心理生理条件下叙述的讲话者的随机语音的语音和不同语言的语音。本方法可广泛应用于刑事调查。基于语音录音的扬声器的识别是通过估计扬声器的第一个录音和第二个校准录音的收敛来进行的。为了进行估计，该方法包括在包含至少三个共振峰的共振峰轨迹的语音信号的第一和第二录音参考片段上选择比较至少两个共振峰频率的值相符的参考片段，基于剩余共振峰频率的值的一致性，比较参考片段的收敛，并且基于所有比较的参考片段的收敛的全局估计来确定所有录音的收敛。

2. 发明申请

WO2004075166A2 BIO-PHONETIC MULTI-PHRASE SPEAKER IDENTITY VERIFICATION 审中-公开
标题翻译：生物多媒体多媒体播放器身份验证
公开(公告)号：WO2004075166A2
公开(公告)日：2004-09-02
申请号：PCT/US2004/002116
申请日：2004-01-26
申请人： SBC PROPERTIES, L.P. , CHANG, Hisao, M.
发明人： CHANG, Hisao, M.
IPC分类号： G10L
CPC分类号： G10L17/26 , G10L17/04 , G10L17/14 , G10L17/22 , G10L17/24
摘要： A speaker identity claim (SIC) utterance is received and recognized. The SIC utterance is compared with a voice profile registered under the SIC, and a first verification decision is based thereon. A first dynamic phrase (FDP) is generated, and a user is prompted to speak same. An FDP utterance is received, and compared with the voice profile registered under the SIC to make a second verification decision. If the second verification decision indicates a high or low confidence level, the speaker identity claim is accepted or rejected, respectively. If the verification decision indicates a medium confidence level, a second dynamic phrase (SDP) is generated, and the user is prompted to speak same. An SDP utterance is received, and compared with the voice profile registered under the SIC to make a third verification decision. The speaker identity claim is accepted or rejected based on the third verification decision.
摘要翻译：发言人身份证明（SIC）话语被接收和认可。将SIC语音与在SIC下注册的语音简档进行比较，并且基于此进行第一验证决定。生成第一个动态短语（FDP），并提示用户说话。接收到FDP话音，并与SIC下注册的语音配置文件进行比较，作出第二个验证决定。如果第二验证决定指示高或低置信水平，说话者身份声明分别被接受或拒绝。如果验证决定指示中等置信水平，则生成第二动态短语（SDP），并且提示用户说话。接收到SDP话音，并与SIC下注册的语音配置文件进行比较，作出第三个验证决定。演讲者身份声明根据第三次验证决定被接受或拒绝。

3. 发明申请

WO99023643A1 MODEL ADAPTATION SYSTEM AND METHOD FOR SPEAKER VERIFICATION 审中-公开
标题翻译：用于演讲者验证的模型适配系统和方法
公开(公告)号：WO99023643A1
公开(公告)日：1999-05-14
申请号：PCT/US1998/023477
申请日：1998-11-03
IPC分类号： G10L15/06 , G10L17/00 , G10L9/06
CPC分类号： G10L15/07 , G10L17/04 , G10L17/14 , G10L17/18 , G10L17/20
摘要： The model adaptation system of the present invention is a speaker verification system that embodies the capability to adapt models learned during the enrollment component to track aging of a user's voice. The system has the advantage of only requiring a single enrollment for recognition models including neural tree networks (22), Gaussian Mixture Models (26), and dynamic time warping (16) or to multiple models (30) (i.e. combinations of neural tree networks (22), Gaussian Mixture Models (26) and Dynamic time warping (30)). Moreover, the present invention can be applied to text-dependent or text-independent systems.
摘要翻译：本发明的模型适应系统是一种扬声器验证系统，其体现了适应在注册组件期间学习的模型以跟踪用户语音的老化的能力。该系统的优点是只需要一个识别模型，包括神经树网络（22），高斯混合模型（26）和动态时间扭曲（16）或多个模型（30）（即神经树网络的组合）（22），高斯混合模型（26）和动态时间扭曲（30））。此外，本发明可以应用于文本依赖或文本无关系统。

4. 发明申请

WO2016044027A1 METHOD AND APPARATUS FOR PERFORMING SPEAKER RECOGNITION 审中-公开
标题翻译：执行扬声器识别的方法和装置
公开(公告)号：WO2016044027A1
公开(公告)日：2016-03-24
申请号：PCT/US2015/049205
申请日：2015-09-09
申请人： NUANCE COMMUNICATIONS, INC.
发明人： ALEY-RAZ, Almong , FARRELL, Kevin, R. , YARON, Oshrit , SCARPATO, Luca
IPC分类号： G10L17/14 , G10L17/24
CPC分类号： G10L17/12 , G10L17/06 , G10L17/14 , G10L17/24
摘要： Embodiments of the present invention perform speaker identification and verification by first prompting a user to speak a phrase that includes a common phrase component and a personal identifier. Then, the embodiments decompose the spoken phrase to locate the personal identifier. Finally, the embodiments identify and verify the user based on the result of the decomposing.
摘要翻译：本发明的实施例通过首先提示用户说出包括公共短语组件和个人标识符的短语来执行说话人识别和验证。然后，实施例分解口头短语以定位个人标识符。最后，实施例基于分解的结果识别和验证用户。

5. 发明申请

WO2014109847A1 SPEAKER VERIFICATION AND IDENTIFICATION USING ARTIFICIAL NEURAL NETWORK-BASED SUB-PHONETIC UNIT DISCRIMINATION 审中-公开
标题翻译：使用基于人工神经网络的子电话歧视的扬声器验证和识别
公开(公告)号：WO2014109847A1
公开(公告)日：2014-07-17
申请号：PCT/US2013/073374
申请日：2013-12-05
申请人： SENSORY, INCORPORATED
发明人： HOSOM, John-Paul , VERMEULEN, Pieter J. , SHAW, Johnathan
IPC分类号： G10L15/16 , G10L15/02 , G10L15/28
CPC分类号： G10L17/14 , G06N3/082 , G06N3/084 , G10L17/04 , G10L17/18
摘要： In one embodiment, a computer system stores speech data for a plurality of speakers, where the speech data includes a plurality of feature vectors and, for each feature vector, an associated sub-phonetic class. The computer system then builds, based on the speech data, an artificial neural network (ANN) for modeling speech of a target speaker in the plurality of speakers, where the ANN is configured to discriminate between instances of sub-phonetic classes uttered by the target speaker and instances of sub-phonetic classes uttered by other speakers in the plurality of speakers.
摘要翻译：在一个实施例中，计算机系统存储用于多个扬声器的语音数据，其中语音数据包括多个特征向量，并且对于每个特征向量，存在相关联的子语音类。计算机系统然后基于语音数据构建用于对多个扬声器中的目标扬声器的语音进行建模的人造神经网络（ANN），其中ANN被配置为区分由目标发出的子语音类别的实例扬声器和多个扬声器中的其他扬声器发出的子语音类的实例。

6. 发明申请

WO2012049368A1 METHOD OF LINGUISTIC PROFILING 审中-公开
标题翻译：语言分析方法
公开(公告)号：WO2012049368A1
公开(公告)日：2012-04-19
申请号：PCT/FI2011/050882
申请日：2011-10-12
申请人： MARTTILA, Annu , PRONOUNCER EUROPE OY
发明人： MARTTILA, Annu
IPC分类号： G10L15/18 , G10L15/06 , G10L15/02 , G10L17/00 , G09B19/06
CPC分类号： G09B5/00 , G09B19/04 , G09B19/06 , G10L17/14
摘要： In the method, in order to define or measure the language proficiency of a person, particularly the degree of flawlessness in the pronunciation, and/or to find out the linguistic background and identity of a person, the person's speech is compared with a selected reference language. This is carried out by applying autocorrelation and/or pattern recognition and/or signal processing and/or other corresponding methods for identifying and registering such sound elements and features that are typical of the reference language and occur repeatedly in the reference language speech sample. On the basis of the obtained linguistic profile of the reference language, corresponding sound elements and features are searched in the speech of said person, and there is calculated how many of the sound elements and features of the reference language linguistic profile the person substitutes with such sound elements or features that deviate from the reference language, and said substitute sound elements and features are defined. For applying the method, there is used a device including a memory unit suited for an electronic recording of speech samples of several different languages, and a computer program that enables the linguistic profiling of several languages and the mutual comparison of several speech samples as well as the registering of the detected differences.
摘要翻译：在该方法中，为了定义或测量一个人的语言能力，特别是发音的完美程度，和/或找出一个人的语言背景和身份，将该人的言语与所选择的参照进行比较语言。这是通过应用自相关和/或模式识别和/或信号处理和/或其他相应的方法来进行的，用于识别和注册参考语言典型的声音元素和特征，并在参考语言语音样本中重复出现。在所获得的参考语言的语言特征的基础上，在所述人的语音中搜索对应的声音元素和特征，并且计算出人们用这种方式替代的参考语言语言概况的声音元素和特征的数量声音元素或特征偏离参考语言，并且所述替代声音元素和特征被定义。为了应用该方法，使用包括适用于几种不同语言的语音样本的电子记录的存储器单元的装置，以及能够进行几种语言的语言分析和几种语音样本的相互比较的计算机程序，以及注册检测到的差异。

7. 发明申请

WO1996041334A1 SPEAKER VERIFICATION SYSTEM 审中-公开
标题翻译：扬声器验证系统
公开(公告)号：WO1996041334A1
公开(公告)日：1996-12-19
申请号：PCT/US1996009260
申请日：1996-06-06
申请人： RUTGERS UNIVERSITY
发明人： RUTGERS UNIVERSITY , MAMMONE, Richard, J. , FARRELL, Kevin , SHARMA, Manish , DEVANG, Naik , ZHANG, Xiaoyu , ASSALEH, Khaled , LIOU, Han-Sheng
IPC分类号： G10L09/00
CPC分类号： G10L17/10 , G10L17/04 , G10L17/14 , G10L17/18 , G10L17/20 , G10L25/03
摘要： The present invention relates to a pattern recognition system (Fig. 1) which uses data fusion to combine data from a plurality of extracted features (60, 61, 62) and a plurality of classifiers (70, 71, 72). Speaker patterns can be accurately verified with the combination of discriminant based and distortion based classifiers. A novel approach using a training set of a "leave one out" data can be used for training the system with a reduced data set (Figs. 7A, 7B, 7C). Extracted features can be improved with a pole filtered method for reducing channel effects (Fig. 11B) and an affine transformation for improving the correlation between training and testing data (Fig. 14).
摘要翻译：本发明涉及使用数据融合来组合来自多个提取特征（60,61,62）和多个分类器（70,71,72）的数据的模式识别系统（图1）。可以使用基于判别式和基于失真的分类器的组合来准确地验证扬声器模式。可以使用使用“保留一个”数据的训练集的新颖方法来用减少的数据集来训练系统（图7A，7B，7C）。通过用于减少信道效应的极点滤波方法（图11B）和用于改善训练和测试数据之间的相关性的仿射变换（图14）可以提高提取的特征。

8. 发明申请

WO2018010683A1 身份向量生成方法、计算机设备和计算机可读存储介质审中-公开
公开(公告)号：WO2018010683A1
公开(公告)日：2018-01-18
申请号：PCT/CN2017/092892
申请日：2017-07-14
申请人：腾讯科技(深圳)有限公司
发明人：李为 , 钱柄桦 , 金星明 , 李科 , 吴富章 , 吴永坚 , 黄飞跃
IPC分类号： G10L17/02
CPC分类号： G10L17/02 , G10L17/14 , G10L17/20
摘要：一种身份向量生成方法，包括：获取待处理语音数据（S302，S402）；从所述待处理语音数据提取相应的声学特征（S304，S404）；对各所述声学特征属于说话人背景模型中每个高斯分布分量的后验概率进行统计得到统计量（S306）；将所述统计量映射到统计量空间获得参考统计量；所述统计量空间根据超过预设语音时长的语音样本所对应的统计量构建而成（S308）；根据统计得到的所述统计量和所述参考统计量确定修正统计量（S310）；及根据所述修正统计量生成身份向量（S312）。该方法能够补偿因待处理语音数据的语音时长过短和语音稀疏的情况下导致的统计量偏估，提高身份向量的身份识别性能。

9. 发明申请

WO2015156798A1 IDENTIFICATION BY SOUND DATA 审中-公开
标题翻译：通过声音数据识别
公开(公告)号：WO2015156798A1
公开(公告)日：2015-10-15
申请号：PCT/US2014/033525
申请日：2014-04-09
申请人： EMPIRE TECHNOLOGY DEVELOPMENT, LLC
发明人： FINE, Kevin, S.
IPC分类号： G10L17/00
CPC分类号： G10L25/54 , G06F17/30766 , G06Q30/0251 , G06Q50/01 , G10L15/30 , G10L17/14 , G10L17/22
摘要： Technologies are generally described for systems, devices and methods effective to identify an individual. In some examples, a microphone may receive sound data such as sound that may be present in a mall. A processor, that may be in communication with the microphone, may determine a name from the sound data. Stated differently, the processor may determine that the name is part of or included in the sound data. The processor may generate a query based on the name and may send the query to a social network database. The processor may receive a response to the query from the social network database and may identify the individual based on the response.
摘要翻译：技术通常描述为有效识别个人的系统，设备和方法。在一些示例中，麦克风可以接收诸如可能存在于商场中的声音的声音数据。可以与麦克风通信的处理器可以根据声音数据确定名称。换句话说，处理器可以确定该名称是声音数据的一部分或包括在声音数据中。处理器可以基于名称生成查询，并且可以将查询发送到社交网络数据库。处理器可以从社交网络数据库接收对查询的响应，并且可以基于响应识别个体。

10. 发明申请

WO2008100971A1 TEXT-DEPENDENT SPEAKER VERIFICATION 审中-公开
标题翻译：文本相关的扬声器验证
公开(公告)号：WO2008100971A1
公开(公告)日：2008-08-21
申请号：PCT/US2008/053769
申请日：2008-02-12
申请人： MICROSOFT CORPORATION
发明人： ZHANG, Zhengyou , SUBRAMAYA, Amarnag
IPC分类号： G10L17/00
CPC分类号： G10L17/24 , G10L17/14
摘要： A text-dependent speaker verification technique that uses a generic speaker-independent speech recognizer for robust speaker verification, and uses the acoustical model of a speaker-independent speech recognizer as a background model. Instead of using a likelihood ratio test (LRT) at the utterance level (e.g., the sentence level), which is typical of most speaker verification systems, the present text-dependent speaker verification technique uses weighted sum of likelihood ratios at the sub-unit level (word, tri-phone, or phone) as well as at the utterance level.
摘要翻译：一种文本相关的扬声器验证技术，其使用一般的与扬声器无关的语音识别器用于强大的扬声器验证，并且使用与扬声器无关的语音识别器的声学模型作为背景模型。现在的文本相关说明者验证技术不是在大多数说话人验证系统的典型的话语级别（例如，句子级别）上使用似然比检验（LRT），而是使用子单元中的似然比加权和水平（单词，三话电话或电话）以及话语水平。

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