公开(公告)号：KR1020100073161A

公开(公告)日：2010-07-01

申请号：KR1020080131755

申请日：2008-12-22

Applicant: 한국전자통신연구원

Inventor： 강점자 ,  이윤근 ,  박전규 ,  정호영 ,  전형배 ,  정훈 ,  이성주 ,  정의석 ,  왕지현 ,  강병옥 ,  박기영 ,  김종진

IPC: G10L15/02 ,  G10L15/08 ,  G10L15/20

CPC classification number: G10L15/187 ,  G10L15/10

Abstract: PURPOSE: An utterance verification method and device for isolated word an NBEST recognition result are provided to enable more reliable voice recognition by displaying the acceptance/refusal or decision failure of voice recognition by measuring inter-phoneme similarity through DTW(Dynamic Time Warping). CONSTITUTION: A pre-processing unit(104) performs feature extraction and end point detection for detecting voice section and noise processing section. An NBEST voice recognition unit(106) perform an NBEST voice recognition through a viterbi speech in consideration of a context-subordinate sound model(26). An NBEST speech verification unit(108) compares the result of an SVM(Support Vector Machine) with the similarity result to measure the similarity of voice recognition result. Therefore, the NBEST speech verification unit displays the acceptance, refusal and decision failure for the voice recognition result.

Abstract translation: 目的：提供一种用于隔离词的话语验证方法和设备，其包括NBEST识别结果，以通过通过DTW（动态时间扭曲）测量语音间相似度来显示语音识别的接受/拒绝或决策失败来实现更可靠的语音识别。 构成：预处理单元（104）执行用于检测语音段和噪声处理部分的特征提取和终点检测。 考虑到上下文从属声音模型（26），NBEST语音识别单元（106）通过维特比语音执行NBEST语音识别。 NBEST语音验证单元（108）将SVM（支持向量机）的结果与相似度结果进行比较，以测量语音识别结果的相似度。 因此，NBEST语音验证单元显示语音识别结果的接受，拒绝和决策失败。

公开(公告)号：KR1020100072746A

公开(公告)日：2010-07-01

申请号：KR1020080131238

申请日：2008-12-22

Applicant: 한국전자통신연구원

Inventor： 정호영 ,  박기영 ,  이성주 ,  강병옥 ,  박전규 ,  정훈 ,  김종진 ,  왕지현 ,  강점자 ,  정의석 ,  전형배 ,  이윤근

IPC: G10L15/20 ,  G10L21/0208 ,  G10L21/0272

Abstract: PURPOSE: A method and an apparatus for reducing a multi channel noise are provided to selectively apply beam-forming method and sound source separating method according to environment condition among multi-channel noises processing based on multi-channel voice recognition environment thereby maximizing noise processing performance. CONSTITUTION: A noise environment monitoring unit(210) grasps the number of sound source and background sound source and relative location information of user voice. According to how many information of the sound source and the relative location information of the background sound source and the user voice, a multi-channel noise processor(220) selects multi-channel noise processing methods among a plurality of multi-channel noises processing modes. The multi-channel noise processor performs noises processing by selected multi-channel noise processing method.

Abstract translation: 目的：提供一种降低多通道噪声的方法和装置，以便根据多声道噪声处理，根据环境条件选择性地应用波束形成方法和声源分离方法，从而使噪声处理性能最大化 。 构成：噪声环境监测单元（210）掌握用户声音的声源和背景声源的数量和相对位置信息。 根据声源的多少信息和背景声源和用户声音的相对位置信息，多声道噪声处理器（220）在多个多声道噪声处理模式中选择多声道噪声处理方法 。 多声道噪声处理器通过选择的多声道噪声处理方法进行噪声处理。

公开(公告)号：KR1020100072731A

公开(公告)日：2010-07-01

申请号：KR1020080131221

申请日：2008-12-22

Applicant: 한국전자통신연구원

Inventor： 왕지현 ,  정의석 ,  강병옥 ,  박전규 ,  강점자 ,  김종진 ,  박기영 ,  이성주 ,  전형배 ,  정호영 ,  정훈 ,  이윤근

IPC: G01C21/34 ,  G01C21/36

CPC classification number: G01C21/3608 ,  G01C21/3611 ,  G01C21/3679 ,  G10L15/26

Abstract: PURPOSE: An apparatus for generating keyword for speech recognition for a navigation device is provided to enable a retrieval service of POI through voice by automatically producing the allomorph of POI which a user can say to a navigation device. CONSTITUTION: An apparatus for generating keyword for speech recognition in navigation device comprises a statistical model studying unit(202) and an allomorph generating unit. The statistical model studying unit analyzes the POI character strings. The statistical model studying unit builds probability value as statistical information. The allomorph generating unit creates the allomorphs on POI name using the statistical information.

Abstract translation: 目的：提供一种用于为导航装置生成用于语音识别的关键词的装置，以通过自动产生用户可以对导航装置说的POI的变体，从而通过语音实现POI的检索服务。 构成：用于在导航装置中生成用于语音识别的关键词的装置，包括统计模型研究单元（202）和变形生成单元。 统计模型研究单位分析POI字符串。 统计模型研究单位建立概率值作为统计信息。 变异生成单元使用统计信息在POI名称上创建变形。

公开(公告)号：KR1020100066916A

公开(公告)日：2010-06-18

申请号：KR1020080125433

申请日：2008-12-10

Applicant: 한국전자통신연구원

Inventor： 박기영 ,  이성주 ,  강병옥 ,  정호영 ,  이윤근 ,  박전규 ,  강점자 ,  정훈 ,  김종진 ,  정의석 ,  전형배 ,  왕지현

IPC: G10L99/00

Abstract: PURPOSE: A method for separating noise from an audio signal is provided to increase performance of sound source separation and increase convergence speed in a weighted learning stage, thereby increasing calculation efficiency. CONSTITUTION: A plurality of microphones records an audio signal that a user speaks and a noise signal. A beam former(20) performs a beam forming process and a blind processing separation procedure for the recorded audio signal and noise signal. The beam former spatially and statistically divides the audio signal and the noise signal. A sound source separator(30) separates the sound source signal and outputs the separated sound source signal.

Abstract translation: 目的：提供一种从音频信号中分离噪声的方法，以增加声源分离的性能，增加加权学习阶段的收敛速度，从而提高计算效率。 构成：多个麦克风记录用户说话的音频信号和噪声信号。 波束成形器（20）对记录的音频信号和噪声信号执行波束形成处理和盲目处理分离程序。 波束形成器在空间和统计学上划分音频信号和噪声信号。 声源分离器（30）分离声源信号并输出​​分离的声源信号。

公开(公告)号：KR1020100026187A

公开(公告)日：2010-03-10

申请号：KR1020080085095

申请日：2008-08-29

Applicant: 한국전자통신연구원

Inventor： 왕지현 ,  정의석 ,  강병옥 ,  이윤근 ,  박기영 ,  이성주 ,  정호영 ,  박전규 ,  강점자 ,  정훈 ,  김종진 ,  전형배

IPC: G10L15/08 ,  G10L15/26 ,  H04N21/438

Abstract: PURPOSE: A voice recognition information generation device, a method thereof, and a broadcast service method thereof are provided to generate a database from allomorph character string, thereby offering a broadcast service according to voice recognition. CONSTITUTION: A voice recognition information generation device includes a prior matching unit(302), a section boundary partition unit(308), a normalization unit(310), and an allomorph generation unit(312). The prior matching unit performs prior matching according to character string information of broadcast data. The section boundary partition unit partitions the section boundary of a character string of which prior matching is performed in order to generate voice recognition target character string data. The normalization unit normalizes generated voice recognition target character string. The allomorph generation unit generates allomorph character string data from normalized voice recognition target character string data.

Abstract translation: 目的：提供一种语音识别信息生成装置，其方法和广播服务方法，以从变形字符串生成数据库，从而根据语音识别提供广播服务。 构成：语音识别信息生成装置包括先验匹配单元（302），区间边界分割单元（308），归一化单元（310）和变形函数生成单元（312）。 先前的匹配单元根据广播数据的字符串信息执行先前的匹配。 区段边界分割单元对执行了先前匹配的字符串的区域边界进行分割，以便生成语音识别目标字符串数据。 归一化单元对生成的语音识别目标字符串进行归一化。 变形生成单元从归一化的语音识别目标字符串数据生成变形字符串数据。

公开(公告)号：KR100930039B1

公开(公告)日：2009-12-07

申请号：KR1020070133217

申请日：2007-12-18

Applicant: 한국전자통신연구원

Inventor： 조훈영 ,  이윤근 ,  정호영 ,  강병옥 ,  강점자 ,  김갑기 ,  이성주 ,  정훈 ,  박전규 ,  전형배

IPC: G10L15/01 ,  G10L15/10 ,  G10L15/20

CPC classification number: G10L15/01

Abstract: An apparatus for evaluating the performance of speech recognition includes a speech database for storing N-number of test speech signals for evaluation. A speech recognizer is located in an actual environment and executes the speech recognition of the test speech signals reproduced using a loud speaker from the speech database in the actual environment to produce speech recognition results. A performance evaluation module evaluates the performance of the speech recognition by comparing correct recognition results answers with the speech recognition results.

Abstract translation: 用于评估语音识别性能的装置包括用于存储N个用于评估的测试语音信号的语音数据库。 语音识别器位于实际环境中，并且在实际环境中从语音数据库执行使用扬声器再现的测试语音信号的语音识别以产生语音识别结果。 性能评估模块通过比较正确的识别结果答案和语音识别结果来评估语音识别的性能。

公开(公告)号：KR1020090065746A

公开(公告)日：2009-06-23

申请号：KR1020070133217

申请日：2007-12-18

Applicant: 한국전자통신연구원

Inventor： 조훈영 ,  이윤근 ,  정호영 ,  강병옥 ,  강점자 ,  김갑기 ,  이성주 ,  정훈 ,  박전규 ,  전형배

IPC: G10L15/01 ,  G10L15/10 ,  G10L15/20

CPC classification number: G10L15/01

Abstract: A device and a method for evaluating performance of a speech recognition engine are provided to require no interference of a person in any noise environment by adjusting an SNR(Signal-to-Noise Ratio) based on free volume control of a speech sound in a speaker. An evaluation speech database(201) stores evaluation speeches. An automatic voice recognition evaluator(203) plays the stored evaluation speech. The automatic voice recognition evaluator transmits an answer list and an audio signal file of evaluation data when voice recognition control for the evaluation data is completed. A speech recognizer(207) recognizes voice, and stores a voice recognition result list and a voice signal file used in voice recognition. A performance evaluation block(209) evaluates performance of a voice recognizer by comparing the answer list and the audio file with the voice recognition result list and the voice signal file.

Abstract translation: 提供了一种用于评估语音识别引擎的性能的装置和方法，用于通过基于扬声器中的语音的自由音量控制来调整SNR（信噪比）来不要求任何噪声环境中的人的干扰 。 评价语音数据库（201）存储评价语句。 自动语音识别评估器（203）播放存储的评估语音。 当评估数据的语音识别控制完成时，自动语音识别评估器发送评估数据的答案列表和音频信号文件。 语音识别器（207）识别语音，并存储语音识别结果列表和用于语音识别的语音信号文件。 性能评估块（209）通过将应答列表和音频文件与语音识别结果列表和语音信号文件进行比较来评估语音识别器的性能。

公开(公告)号：KR1020090061566A

公开(公告)日：2009-06-16

申请号：KR1020080088318

申请日：2008-09-08

Applicant: 한국전자통신연구원

Inventor： 조훈영 ,  이윤근 ,  강점자 ,  강병옥 ,  김갑기 ,  이성주 ,  정호영 ,  정훈 ,  박전규 ,  전형배

IPC: G10L15/20

Abstract: A microphone array-based voice recognition system and a target voice extracting method in the system are provided to automatically find out one target voice uttered for voice recognition by using an HMM(Hidden Markov Model) and a GMM(Gaussian Mixture Model), thereby obtaining a higher recognition rate even in case of noise existence. A signal separator(110) separates mixed signals individually inputted through plural microphones into sound source signals through independent component analysis. A target voice extractor(120) extracts one target voice uttered for voice recognition among the separated sound source signals. A voice recognizer(130) recognizes a desired voice through the extracted target voice. An additional information unit transmits additional information used for the extraction of the target voice to the target voice extractor.

Abstract translation: 提供了一种基于麦克风阵列的语音识别系统和系统中的目标语音提取方法，通过使用HMM（隐马尔可夫模型）和GMM（高斯混合模型）自动找出发出语音识别的一个目标语音，从而获得 甚至在噪声存在的情况下也具有更高的识别率。 信号分离器（110）通过独立分量分析将通过多个麦克风分别输入的混合信号分离成声源信号。 目标语音提取器（120）在分离的声源信号之间提取用于语音识别发出的一个目标语音。 语音识别器（130）通过所提取的目标语音来识别期望的语音。 附加信息单元将用于提取目标语音的附加信息发送到目标语音提取器。

公开(公告)号：KR1020070061216A

公开(公告)日：2007-06-13

申请号：KR1020060066884

申请日：2006-07-18

Applicant: 한국전자통신연구원

Inventor： 이성주

IPC: G10L21/0216 ,  G10L15/20 ,  G10L19/02

Abstract: A speech enhancement system using a GMM(Gaussian Mixture Model) is provided to check the existence of dynamic noise in an input signal using a GMM, estimate frequency spectrum characteristic of the dynamic noise and remove the dynamic noise from the input signal to enhance speech quality. A speech enhancement system includes a speech enhancement filter(270), a frequency spectrum estimator(210), a static noise estimator(230), a dynamic noise estimator(240), a noise characteristic estimator(250), and a filter frequency characteristic controller(260). The speech enhancement filter has filtering characteristic which is controlled in real time in order to enhance speech quality of an input signal. The frequency spectrum estimator analyzes the input signal on a frequency band. The static noise estimator calculates a degree of static noise in the output signal of the frequency spectrum estimator according to predetermined static noise frequency characteristic. The dynamic noise estimator calculates a degree of dynamic noise in the output signal of the frequency spectrum estimator according to a dynamic noise GMM. The noise characteristic estimator estimates noise characteristic of the input signal from information on the degree of static noise and the degree of dynamic noise. The filter frequency characteristic controller controls the filtering characteristic of the speech enhancement filter according to the noise characteristic of the input signal.

Abstract translation: 提供使用GMM（高斯混合模型）的语音增强系统，以使用GMM来检查输入信号中的动态噪声的存在，估计动态噪声的频谱特性，并从输入信号中去除动态噪声以增强语音质量 。 语音增强系统包括语音增强滤波器（270），频谱估计器（210），静态噪声估计器（230），动态噪声估计器（240），噪声特征估计器（250）和滤波器频率特征 控制器（260）。 语音增强滤波器具有实时控制的滤波特性，以便提高输入信号的语音质量。 频谱估计器分析频带上的输入信号。 静态噪声估计器根据预定的静态噪声频率特性来计算频谱估计器的输出信号中的静态噪声的程度。 动态噪声估计器根据动态噪声GMM计算频谱估计器的输出信号中的动态噪声程度。 噪声特征估计器根据关于静态噪声程度和动态噪声程度的信息估计输入信号的噪声特性。 滤波器频率特性控制器根据输入信号的噪声特性来控制语音增强滤波器的滤波特性。

公开(公告)号：KR1020050036301A

公开(公告)日：2005-04-20

申请号：KR1020030071935

申请日：2003-10-15

Applicant: 한국전자통신연구원

Inventor： 유미선 ,  김승희 ,  이성주

IPC: G10L15/08

Abstract: 본 발명은 피치와 MFCC를 이용한 성별식별 장치 및 방법에 관한 것으로, 임의의 음성을 입력받아 발성한 화자의 성별을 자동적으로 식별함으로써, 음성인식이나 화자식별, 기타 성별에 따라 다른 서비스를 제공할 필요가 있는 응용프로그램의 전처리 모듈에 사용될 수 있다.
특히, 본 발명의 특징은 다수의 화자 음성이 남녀로 분리되어 저장된 훈련음성 데이터베이스, 상기 훈련음성 데이터베이스를 토대로 남녀별 피치값 분포와 기준 피치값이 저장된 남녀 표준 피치값 저장부, 상기 훈련음성 데이터베이스를 토대로 남녀별 GMM(Gaussian Mixture Model)이 저장된 남녀 GMM 저장부와, 화자의 음성이 입력되는 입력음성 데이터부, 상기 입력음성 데이터부로 입력된 음성에서 유성음 구간으로부터 측정된 피치값과 상기 남녀 표준 피치값 저장부에 저장된 피치정보를 비교하여 남녀 성별을 추정하는 제 1단계 화자성별 식별수단 및 상기 입력음성 데이터부로 입력된 음성파일로부터 추출된 MFCC(Mel-Frequency Cepstral Coefficients)값과 상기 남녀 GMM 저장부에 저장된 GMM 정보를 이용하여 남녀 성별을 추정하는 제 2단계 화자성별 식별수단으로 구성 되는 것을 특징으로 한다.