-
公开(公告)号：KR1020140023852A
公开(公告)日：2014-02-27
申请号：KR1020130064263
申请日：2013-06-04
Applicant: 한국전자통신연구원
IPC: H04N21/441 , H04N21/45
CPC classification number: H04N21/4312 , H04N21/25875 , H04N21/274 , H04N21/4753 , H04N21/4858
Abstract: Provided are an apparatus and method for providing a personalized home screen which provide various personalized home screens for a plurality of users who use the same smart TV and provide the same home screen for a different smart TV using personal information stored in a cloud server. The apparatus for providing a personalized home screen according to the present invention comprises: a user identification unit for receiving user identification information from a user, and transferring the user identification information; and a home screen processing unit for comparing the user identification information received from the user to personal information of multiple users, and outputting a personalized home screen on a screen based on the identified user personal information. As a result, user can conveniently use a smart TV by being provided an appropriate home screen for the user, and can be provided the same home screen although the user use a different smart TV. [Reference numerals] (110) User identification unit; (120) Input unit; (130) Home screen processing unit; (140) Cloud server; (AA) User personal information request; (BB) User personal information; (CC) User identification information; (DD) Control signal
Abstract translation: 提供了一种用于提供个性化主屏幕的装置和方法，其为使用相同智能电视的多个用户提供各种个性化主屏幕，并且使用存储在云服务器中的个人信息为不同智能电视机提供相同的主屏幕。 根据本发明的用于提供个性化主屏幕的装置包括：用户识别单元，用于从用户接收用户识别信息，并传送用户识别信息; 以及家庭屏幕处理单元，用于将从用户接收的用户识别信息与多个用户的个人信息进行比较，并且基于所识别的用户个人信息在屏幕上输出个性化主屏幕。 结果，用户可以通过为用户提供适当的主屏幕来方便地使用智能电视，并且可以提供相同的主屏幕，尽管用户使用不同的智能电视。 （附图标记）（110）用户识别单元; （120）输入单元; （130）主屏处理单元; （140）云服务器; （AA）用户个人信息请求; （BB）用户个人信息; （CC）用户识别信息; （DD）控制信号
-
公开(公告)号：KR1020130059476A
公开(公告)日：2013-06-07
申请号：KR1020110125405
申请日：2011-11-28
Applicant: 한국전자통신연구원
IPC: G10L15/08
CPC classification number: G10L15/083 , G10L15/187 , G10L15/08 , G10L2015/081
Abstract: PURPOSE: A search space generating method for voice recognition and a system thereof are provided to improve an accuracy of a voice recognition by adding 'a pronunciation heat which is generated by a pronunciation conversion between recognition units' to a search space. CONSTITUTION: A WFST[Weighted Finite State Transducer] coupling unit generates a WFST L·G by a coupling of a WFST G[WFST Grammar] and a WFST L[WFST pronunciation Library] and generates a WFST L'·L·G by a coupling of a WFST L'[WFST pronunciation conversion] and the WFST L·G(310,320). The WFST coupling unit generates a WFST C·L'·L·G by a coupling of a WFST context[WFST C] and the WFST L'·L·G and generates a WFST H·C·L'·L·G by a coupling of a WFST H[WFST Hidden Markov model] and the WFST C·L'·L·G(330,340). A WFST optimization unit optimizes the WFST H·C·L'·L·G(350). [Reference numerals] (310) WFST G and WFST L combination; (320) WFST L' and WFST L·G combination; (330) WFST C and WFST L'·L·G combination; (340) WFST H and WFST C·L'·L·G combination; (350) Optimization; (AA) Start; (BB) End
Abstract translation: 目的：提供一种用于语音识别的搜索空间生成方法及其系统，以通过将由识别单元之间的语音转换产生的发音热量添加到搜索空间来提高语音识别的精度。 构成：WFST [加权有限状态传感器]耦合单元通过WFST G [WFST语法]和WFST L [WFST发音库]的耦合产生WFST L·G，并通过一个WFST L'·L·G生成WFST L' WFST L'[WFST发音转换]与WFST L·G（310,320）的耦合。 WFST耦合单元通过WFST上下文[WFST C]和WFST L'·L·G的耦合产生WFST C·L'·L·G，并通过下式产生WFST H·C·L'·L·G WFST H [WFST隐马尔可夫模型]和WFST C·L'·L·G（330,340）的耦合。 WFST优化单元优化WFST H·C·L'·L·G（350）。 （参考号）（310）WFST G和WFST L组合; （320）WFST L'和WFST L·G组合; （330）WFST C和WFST L'·L·G组合; （340）WFST H和WFST C·L'·L·G组合; （350）优化; （AA）开始; （BB）结束
-
公开(公告)号：KR1020120056086A
公开(公告)日：2012-06-01
申请号：KR1020100117611
申请日：2010-11-24
Applicant: 한국전자통신연구원
CPC classification number: G10L15/14 , G10L15/26 , G10L19/038
Abstract: PURPOSE: An acoustic model adapting method and a voice recognizing device using the same are provided to eliminate a re-study burden of a user about a quantized acoustic model by an embedded voice recognizing machine. CONSTITUTION: An extracting unit(110) extracts features from a waveform corresponding to a voice. The extracting unit generates quantized data. A probability measuring unit(120) applies the quantized data, an adapted network, and a quantized acoustic model to fixed point-applied high-speed computation. The probability measuring unit calculates Gaussian occupancy probability. An adaption unit(130) updates the acoustic model. A voice recognizing unit(150) recognizes the extracted features using the updated acoustic model.
Abstract translation: 目的：提供一种声学模型适应方法和使用其的语音识别装置，以消除用户通过嵌入式语音识别机器对量化声学模型的重新学习负担。 构成：提取单元（110）从对应于声音的波形中提取特征。 提取单元生成量化数据。 概率测量单元（120）将量化数据，适应网络和量化声学模型应用于固定点施加的高速计算。 概率测量单元计算高斯占用概率。 适应单元（130）更新声学模型。 语音识别单元（150）使用更新的声学模型识别所提取的特征。
-
公开(公告)号：KR101027518B1
公开(公告)日：2011-04-06
申请号：KR1020080131764
申请日：2008-12-22
Applicant: 한국전자통신연구원
IPC: G10L15/193 , G06F17/27 , G06F17/28
Abstract: 본 발명은 가중 유한 상태 변환기(Weighted Finite State Transducer)에서, 입력 기호가 널(null)인 천이, 예컨대 ε(epsilon ; 엡실론) 입력 천이를 제거하는 방법에 관한 것이다. 가중 유한 상태 변환기의 등가성(equivalency)을 유지하면서, 각 상태(state, 또는 노드(node))의 종류에 따라 ε 입력 천이 제거 방법을 달리하여 가중 유한 상태 변환기의 크기를 최소화하는 방향으로 ε 입력 천이를 제거한다.
가중 유한 상태 변환기, 널 심볼 천이-
公开(公告)号：KR1020100138520A
公开(公告)日：2010-12-31
申请号：KR1020090057093
申请日：2009-06-25
Applicant: 한국전자통신연구원
IPC: G10L15/197 , G10L15/14 , G10L15/06
Abstract: PURPOSE: A speech recognition apparatus and a method thereof are provided to reduce error of remote speech recognition. CONSTITUTION: A syntax analyzing unit(23) analyzes syntax based on a morpheme word class to generate a hierarchical structure. A hierarchical word list generating unit(24) generates a word list by a hierarchy of a recognition word using the hierarchical structure. A hierarchical n-gram applying unit(25) generates a hierarchical n-gram score of the word list by a hierarchy. A calculation unit(27) adds the hierarchical n-gram score to sound and language model probability to generate a speech recognition score of the recognition word.
Abstract translation: 目的：提供语音识别装置及其方法，以减少远程语音识别的误差。 构成：语法分析单元（23）基于语素词类分析语法以生成层次结构。 分级词列表生成单元（24）使用层次结构通过识别词的层次来生成单词列表。 分层的n-gram应用单元（25）通过层级产生单词列表的分层n-gram分数。 计算单元（27）将分级n-gram分数与声音和语言模型概率相加，以产生识别词的语音识别分数。
-
Patent Agency Ranking
公开(公告)号：KR1020100073170A
公开(公告)日：2010-07-01
申请号：KR1020080131764
申请日：2008-12-22
Applicant: 한국전자통신연구원
IPC: G10L15/193 , G06F17/27 , G06F17/28
Abstract: PURPOSE: A null symbol transition removal method in weighting finite state transducer is provided to remove all ε input transitions except a loop formed by ε input transitions, thereby minimizing the size of the weighting finite state transducer according to a form of input transitions of the whole weighted finite state transducer. CONSTITUTION: An input transition search unit(100) searches null symbol transition while searching the whole weighted finite state transducer, for example searching ε input transition. An input transition determination unit(102) determines type of the ε input transition. An input transition removing unit(104) removes the ε input transition according to type of the determined ε input transition. If an input symbol is null, forward direction processing suitable for node type is performed. If the input symbol is not null, the next transition is performed.
Abstract translation: 目的：提供加权有限状态传感器中的零符号跃迁去除方法，以除去除了ε输入跃迁形成的环之外的所有ε输入跃迁，从而根据整体输入转换的形式最小化加权有限状态换能器的大小 加权有限状态传感器。 构成：输入转换搜索单元（100）在搜索整个加权有限状态换能器时搜索零符号转换，例如搜索ε输入转换。 输入转换确定单元（102）确定ε输入转换的类型。 输入转换去除单元（104）根据确定的ε输入转换的类型去除ε输入转变。 如果输入符号为空，则执行适合于节点类型的正向处理。 如果输入符号不为空，则执行下一个转换。
-
公开(公告)号：KR100925479B1
公开(公告)日：2009-11-06
申请号：KR1020070095540
申请日：2007-09-19
Applicant: 한국전자통신연구원
CPC classification number: G10L15/187 , G10L2015/025
Abstract: 본 발명은 음성 인식 방법 및 장치에 관한 것으로, 음소 인식된 음소열에 대한 신뢰도를 계산하고 이를 이용하여 음성 인식 성능을 향상시키기 위한 방법 및 장치를 제공한다. 이를 위하여, 본 발명에 따른 음성 인식 방법은, 음성으로 입력된 문자열에 포함된 음소 간의 경계를 결정함으로써 각 음소 구간을 검출하는 단계; 상기 검출된 각 음소 구간이 나타내는 음소가 미리 정의된 음소 모델에 속하는 각 음소일 확률에 따른 신뢰도를 계산하는 단계; 상기 계산된 신뢰도 및 미리 훈련하여 저장된 음소 인식 확률 분포를 기반으로 상기 문자열에 대한 음소 정렬 비용을 계산하는 단계; 및 상기 계산된 음소 정렬 비용을 기반으로 음소 정렬을 수행함으로써 상기 입력된 문자열을 음성 인식하는 단계를 포함함으로써, 음소 인식된 음소열에 대한 신뢰도를 계산하고 이를 이용하여 음성 인식 성능을 향상시킬 수 있는 이점이 있다.
음성 인식, 유사도, 확률, 신뢰도-
公开(公告)号：KR1020090030166A
公开(公告)日：2009-03-24
申请号：KR1020070095540
申请日：2007-09-19
Applicant: 한국전자통신연구원
CPC classification number: G10L15/187 , G10L2015/025
Abstract: A voice recognition method and an apparatus thereof are provided to calculate the reliability on a grapheme string of recognized grapheme, and improve the voice recognition performance by using the reliability. A phoneme interval detecting unit(404) analyzes voice feature data outputted from a voice feature extracting unit(402), and then determines the boundary between phonemes to detect a phoneme interval. A reliability decision unit(406) compares the pattern of the detected phoneme intervals with phonemes belonging to a pre-defined phoneme model(416) to calculate the similarity. A reliability-based phoneme error model(418) stores the phoneme recognition probability distribution which is obtained by training whether an inputted voice phoneme is recognized as any kind of phoneme, in advance.
Abstract translation: 提供语音识别方法及其装置，以计算识别的图形的字母串的可靠性，并通过使用可靠性来提高语音识别性能。 音素间隔检测单元（404）分析从语音特征提取单元（402）输出的语音特征数据，然后确定音素之间的边界以检测音素间隔。 可靠性决定单元406将所检测的音素间隔的模式与属于预先定义的音素模型（416）的音素进行比较，以计算相似度。 基于可靠性的音素误差模型（418）存储通过训练将输入的语音音素是否被识别为任何种类的音素而获得的音素识别概率分布。
-
公开(公告)号：KR100842754B1
公开(公告)日：2008-07-01
申请号：KR1020060124556
申请日：2006-12-08
Applicant: 한국전자통신연구원
Inventor： 김승희
Abstract: 본 발명은 음성 신호를 수신하여 음성 특징을 추출하는 단계, 추출된 음성 특징에 상응하여 적어도 하나의 조음 특징을 추출하는 단계, 추출된 조음 특징에 상응하여 음소별 조음특징점수를 계산하는 단계 및 음소별 조음특징점수 및 추출된 음성 특징의 관측 확률 값에 상응하여 관측 점수를 계산하는 단계를 포함하되, 음소별 조음특징점수는 적어도 하나의 조음 특징에 상응하는 적어도 하나의 음소를 검색하고, 검색된 적어도 하나의 음소 각각에 상응하는 조음 특징의 신뢰도를 가중합하여 계산하는 것을 특징으로 하는 음성 인식 방법.
음성 인식, 조음 특징, 신뢰도-
公开(公告)号：KR100835985B1
公开(公告)日：2008-06-09
申请号：KR1020060124206
申请日：2006-12-08
Applicant: 한국전자통신연구원
Abstract: An apparatus and a method for recognizing continuous speech using search network limitation based on keyword recognition are provided to recognize keywords capable of limiting the size of a keyword search network, and perform speech recognition through keyword-associated search networks, thereby improving speech recognition speed and performance. An apparatus for recognizing continuous speech comprises a speech property extractor(110), a keyword recognizer(120), and a keyword-based speech recognizer(140). The speech property extractor extracts a specific vector from an input speech signal. The keyword recognizer recognizes and extracts keywords through a keyword search network(130) configured by centering on the keyword capable of limiting the size of the keyword search network based on the specific vector of the speech signal. The keyword-based speech recognizer configures n keyword-associated search networks(150) with a limited search space on the basis of the extracted keywords, and performs the speech recognition based on the keywords through the n keyword-associated search networks. The keyword recognizer extracts non-keywords from the specific vector of the speech signal through a filler model(160) to extract only keywords, and recognizes and extracts the keywords capable of limiting the size of the keyword search network out of the keywords extracted through the keyword search network.
Abstract translation: 提供了一种基于关键词识别的搜索网络限制识别连续语音的装置和方法，以识别能够限制关键词搜索网络的大小的关键词，并通过关键字关联搜索网络进行语音识别，从而提高语音识别速度， 性能。 用于识别连续语音的装置包括语音特征提取器（110），关键字识别器（120）和基于关键字的语音识别器（140）。 语音特征提取器从输入语音信号中提取特定向量。 关键字识别器通过关键词搜索网络（130）来识别和提取关键词，该关键字搜索网络（130）基于能够基于语音信号的特定向量来限制关键字搜索网络的大小的关键词来配置。 基于关键字的语音识别器基于所提取的关键词来配置具有有限搜索空间的n个关键字关联的搜索网络（150），并且通过n个关键字关联的搜索网络基于关键字执行语音识别。 关键词识别器通过填充模型（160）从填充模型（160）从语音信号的特定向量中提取非关键字，以仅提取关键字，并且识别和提取能够限制关键字搜索网络的大小的关键字，从通过 关键词搜索网络。
-
-
-
-
-
-
-
-
-
Search Results
187
records
(took 0.032 seconds)
