-
公开(公告)号：KR1020070090642A
公开(公告)日：2007-09-06
申请号：KR1020060020600
申请日：2006-03-03
Applicant: 삼성전자주식회사
CPC classification number: G06F17/2785 , G06F17/2705
Abstract: An apparatus and a method for providing a voice dialogue service are provided to analyze user's uttered contents including a natural language by using semantic slot recognition, thereby more exactly recognizing the uttered contents of the user irrespective of an utterance type of the user. A language analysis module(220) tags a POS(Part Of Speech) to each word included in a sentence recorded in a predetermined text, performs syntax analysis of the sentence by classifying meaning of each word, and generates one or more semantic frame corresponding to the sentence according to the syntax analysis result. A dialog managing module(230) analyzes intention of the sentence corresponding to the each semantic frame, and generates a system response corresponding to the sentence intention by selecting predetermined sentence intention according to whether an action corresponding to each sentence intention is performable.
Abstract translation: 提供了一种用于提供语音对话服务的装置和方法，以通过使用语义时隙识别来分析包括自然语言的用户的发声内容，从而更准确地识别用户的发声内容，而不管用户的话语类型。 语言分析模块（220）将POS（部分语音）标记到在预定文本中记录的句子中包含的每个单词，通过对每个单词的含义进行分类来执行句子的语法分析，并且生成一个或多个对应于 句子根据语法分析结果。 对话管理模块（230）分析与每个语义帧相对应的句子的意图，并且根据与每个句子意图相对应的动作是否可执行，通过选择预定句子意图来生成与句子意图相对应的系统响应。
-
公开(公告)号：KR100703696B1
公开(公告)日：2007-04-05
申请号：KR1020050011426
申请日：2005-02-07
Applicant: 삼성전자주식회사
Abstract: 본 발명은 제어 명령 인식 방법 및 이를 이용한 제어 장치에 관한 것이다.
본 발명의 실시예에 따른 제어 명령 인식 방법은 사용자로부터 입력 정보를 입력 받는 단계, 커맨드 포커스가 설정된 제어 대상이 있는 경우 소정의 참조 정보를 참조하여 커맨드 포커스가 설정된 제어 대상 및 입력 정보에 매핑되는 제어 명령을 추출하는 단계, 및 추출된 제어 명령에 따르는 제어 신호를 출력하는 단계를 포함한다.
본 발명에 따르면 음성과 버튼을 복합적으로 사용하여 입력되는 제어 명령에 대한 인식율을 높일 수 있다.
음성 인식, 원격 제어, 제어 장치-
公开(公告)号：KR1020040042719A
公开(公告)日：2004-05-20
申请号：KR1020020071306
申请日：2002-11-15
Applicant: 삼성전자주식회사
IPC: G10L13/00
CPC classification number: G10L13/10
Abstract: PURPOSE: A text-to-speech conversion system and method having a function of providing additional information are provided to enable smooth communication through text-to-speech conversion by providing specific words as emphasized words. CONSTITUTION: A text-to-speech conversion system includes a speech synthesizing module(100), an emphasized word selecting module(300), and a display module(900). The text-to-speech conversion system further includes an information type decision module(500) and a structurization module(700). The speech synthesizing module synthesizes speech from input text data with reference to language analysis data and speech synthesis result analysis data. The emphasized word selecting module selects an emphasized word from the input text data using speech synthesis analysis data obtained from the speech synthesizing module. The information type decision module decides an information type of the input text data using the speech synthesis analysis data and generates sentence pattern information.
Abstract translation: 目的：提供具有提供附加信息功能的文本到语音转换系统和方法，以通过提供特定词作为强调词来实现通过文本到语音转换的平滑通信。 构成：文本到语音转换系统包括语音合成模块（100），强调词选择模块（300）和显示模块（900）。 文本转语音转换系统还包括信息类型判定模块（500）和结构化模块（700）。 语音合成模块参考语言分析数据和语音合成结果分析数据，从输入文本数据中合成语音。 强调词选择模块使用从语音合成模块获得的语音合成分析数据，从输入文本数据中选择一个强调词。 信息类型决定模块使用语音合成分析数据来决定输入文本数据的信息类型，并生成句子模式信息。
-
-
公开(公告)号：KR101644015B1
公开(公告)日：2016-08-01
申请号：KR1020090115914
申请日：2009-11-27
Applicant: 삼성전자주식회사
CPC classification number: G10L15/22 , G10L17/10 , G10L2015/223 , G10L2015/227
Abstract: 시스템과다수사용자간의대화인터페이스장치가개시된다. 본발명의일 양상에따른대화인터페이스장치는, 등록된사용자의음성정보인지여부, 등록된사용자가시스템을주목하고있는상태에서발화가이루어진것인지여부, 및등록된사용자의음성정보가시스템에게유의미한지여부에따라발화자의발화정보를시스템에게전달하거나차단한다. 따라서등록된사용자가시스템을주목한상태에서의미있는발성을한 경우에만인터페이싱을하는것이가능하다.
-
Patent Agency Ranking
公开(公告)号：KR1020120119885A
公开(公告)日：2012-10-31
申请号：KR1020120103001
申请日：2012-09-17
Applicant: 삼성전자주식회사
CPC classification number: Y02D10/45
Abstract: PURPOSE: A content category method and a system thereof are provided to reduce power consumption of a mobile terminal according to web connection by utilizing history information for mobile information search. CONSTITUTION: Information related to search of content information is extracted through a network(320). The content information is classified by using the extracted information(360). The extracted information is query information, link information or content information used for searching for content information. The extracted information is parsed. The content information is classified based on a parsing result. [Reference numerals] (310) Web contents download; (320) Query extraction; (321) Query parsing; (322) Expansion to synonym set; (330) Is it web page?; (331) Web page passing; (332) Text information extraction; (333) Is it text file?; (334) Text information extraction; (335) Meta-date extraction; (336) Contents indexing; (340) Change contents file name as query; (350) Construction of query cash and contents DB; (360) Contents classification; (AA) Start; (BB,DD) Yes; (CC,EE) No; (FF) End
Abstract translation: 目的：提供一种内容分类方法及其系统，以通过利用用于移动信息搜索的历史信息来减少根据web连接的移动终端的功耗。 构成：通过网络（320）提取与搜索内容信息有关的信息。 通过使用提取的信息对内容信息进行分类（360）。 所提取的信息是用于搜索内容信息的查询信息，链接信息或内容信息。 提取的信息被解析。 基于解析结果对内容信息进行分类。 （附图标记）（310）Web内容下载; （320）查询提取; （321）查询解析; （322）扩展到同义词集; （330）是网页吗？ （331）网页通过; （332）文本信息提取; （333）是文本文件吗？ （334）文本信息提取; （335）元数据提取; （336）目录索引; （340）将内容文件名更改为查询; （350）查询现金和内容DB的构建; （360）内容分类; （AA）开始; （BB，DD）是; （CC，EE）否; （FF）结束
-
公开(公告)号：KR1020120038198A
公开(公告)日：2012-04-23
申请号：KR1020100099828
申请日：2010-10-13
Applicant: 삼성전자주식회사
CPC classification number: G10L15/197
Abstract: PURPOSE: A voice recognition device and a method thereof are provided to select a language model with high points for each word of a word string from a forward language model and a backward language model and calculate a bidirectional language model point. CONSTITUTION: The first voice recognition unit(110) generates word lattice information through voice recognition for an input voice. A word string generating unit(120) generates at least one word string from the word lattice information. A language model point calculating unit(130) calculates a bidirectional language model point for each word of the word string by using a forward language model(160) and a backward language model. A sentence output unit(140) outputs a word string with high points as the voice recognition result.
Abstract translation: 目的：提供一种语音识别装置及其方法，用于从前向语言模型和反向语言模型中选择具有高点对于字串的每个单词的语言模型，并计算双向语言模型点。 构成：第一语音识别单元（110）通过用于输入语音的语音识别产生字格信息。 字串生成单元（120）从字格信息生成至少一个字串。 语言模型点计算单元（130）通过使用前向语言模型（160）和反向语言模型来计算单词串的每个单词的双向语言模型点。 句子输出单元（140）输出具有高点的字串作为语音识别结果。
-
公开(公告)号：KR1020110119478A
公开(公告)日：2011-11-02
申请号：KR1020100039217
申请日：2010-04-27
Applicant: 삼성전자주식회사
Abstract: PURPOSE: A voice recognition apparatus and voice recognition method are provided to improve the performance of voice recognition by supplementing a language model which includes qualified data. CONSTITUTION: A first voice recognition unit(110) detects first word lattice information through continous word recognition about input voice. A second word lattice information detecting unit(120) detects second word lattice information. A merging unit(130) creates integrated lattice information through the first word lattice information and the second word lattice information. A second voice recognition unit(140) creates a sentence about the input voice using second voice recognition through the integrated lattice information.
Abstract translation: 目的：提供语音识别装置和语音识别方法，通过补充包含合格数据的语言模型来提高语音识别的性能。 构成：第一语音识别单元（110）通过关于输入语音的连续字识别来检测第一字格信息。 第二字格信息检测单元（120）检测第二字格信息。 合并单元（130）通过第一单词格点信息和第二单词格点信息产生综合格点信息。 第二语音识别单元（140）通过集成的格子信息使用第二语音识别创建关于输入语音的句子。
-
公开(公告)号：KR1020110036385A
公开(公告)日：2011-04-07
申请号：KR1020090094019
申请日：2009-10-01
Applicant: 삼성전자주식회사
CPC classification number: G06F17/279 , G06F17/271 , G10L15/1815
Abstract: PURPOSE: A user purpose analysis apparatus and method thereof are provided to prevent the output of a voice recognition result from a sentence having a voice recognition function. CONSTITUTION: A phrase spotting unit(220) performs phrase spotting operation. An available sentence determining unit(230) determines whether at least one sentence is a grammatically effective sentence. The sentence in which a valid sentence determining unit is not effective is filtered. An intension inference unit(240) generates an intension analysis result about a sentence that is determined as an effective sentence.
Abstract translation: 目的：提供用户目的分析装置及其方法，以防止具有语音识别功能的语音的语音识别结果的输出。 构成：短语发现单元（220）执行短语识别操作。 可用句子确定单元（230）确定至少一个句子是否是语法上有效的句子。 过滤有效句子确定单元无效的句子。 推理单元（240）产生关于被确定为有效句子的句子的意图分析结果。
-
公开(公告)号：KR1020110009614A
公开(公告)日：2011-01-28
申请号：KR1020100036031
申请日：2010-04-19
Applicant: 삼성전자주식회사
CPC classification number: G06F9/44505 , G06F3/016 , G06F17/3025 , G06F17/30684 , G06F17/30731 , G06F2203/014
Abstract: PURPOSE: An apparatus for predicting an intention of a user using multi modal information and a method thereof are provided to predict part of an intention of a user through user motion recognition and predict the intention of the user based on analysis of multi modal information by the predicted intention of the user, thereby accurately guessing the intention of the user. CONSTITUTION: The first predicting unit(220) uses motion information to predict part of an intention of a user. The second predicting unit(230) uses the part of the intention of the user and multimodal information inputted from a multi modal sensor to predict an intention of the user. The first predicting unit generates a control signal for executing a predicted operation of the second predicting unit. The control signal controls an operation of the multi modal sensor.
Abstract translation: 目的：提供一种用于使用多模式信息预测用户意图的装置及其方法，用于通过用户运动识别来预测用户的意图的一部分，并且基于多模态信息的分析来预测用户的意图 用户的预期意图，从而准确地猜测用户的意图。 构成：第一预测单元（220）使用运动信息来预测用户的意图的一部分。 第二预测单元（230）使用用户的意图的一部分和从多模态传感器输入的多模态信息来预测用户的意图。 第一预测单元生成用于执行第二预测单元的预测操作的控制信号。 控制信号控制多模式传感器的操作。
-
-
-
-
-
-
-
-
-
Search Results
74
records
(took 0.026 seconds)
