-
公开(公告)号：KR101560834B1
公开(公告)日：2015-10-15
申请号：KR1020090013492
申请日：2009-02-18
Applicant: 삼성전자주식회사 , 연세대학교 산학협력단
Abstract: 본발명은감정을인식하는장치및 방법에관한것으로, 특히음성신호를이용하여감정을인식하는장치에있어서, 입력된발화음으로부터피치궤도와에너지궤도중 적어도하나를비언어적정보로검출하는비언어적정보검출부; 상기검출된피치궤도와에너지궤도중 적어도하나로부터특정구간을검출하고, 상기검출된특정구간으로부터비언어적파라미터를추출하는비언어적파라미터추출부; 및상기추출된비언어적파라미터를이용하여상기발화음에대해인식되는감정상태를판단하는판단부를포함한다.
-
公开(公告)号：KR101336203B1
公开(公告)日：2013-12-05
申请号：KR1020070097898
申请日：2007-09-28
Applicant: 삼성전자주식회사
Inventor： 유성엽
Abstract: 본 발명은 전자기기에서 음성 검출 방법 및 장치에 관한 것으로서, 입력 신호 프레임에 대한 에너지 및 스펙트럼을 측정하는 과정과, 상기 입력 신호에서 소정 수의 초기 프레임에 대한 에너지 평균값 및 스펙트럼 평균값을 산출하는 과정과, 상기 스펙트럼 평균값과 상기 소정 수 이후에 입력된 신호 프레임의 스펙트럼을 이용하여 스펙트럼 차이값을 산출하는 과정과, 상기 에너지 평균값과 상기 소정 수 이후에 입력된 신호 프레임의 에너지값의 비율을 기반으로 신호대 잡음 에너지 비율(Signal to Noise Ratio: 이하 'SNR'이라 칭함)을 산출하는 과정과, 상기 SNR에 따라 동적으로 변화하는 문턱값과 상기 스펙트럼 차이값을 비교하여 상기 입력 신호에서 음성을 검출하는 과정을 포함하여 시간과 공간에 따라 다양하게 변화하는 잡음이 존재하는 상황에서도 � �성 구간과 묵음 구간을 명확하게 검출할 수 있는 효과가 있다.
음성 검출, 음성 검출기, 묵음 구간, 음성 구간-
公开(公告)号：KR1020100094182A
公开(公告)日：2010-08-26
申请号：KR1020090013492
申请日：2009-02-18
Applicant: 삼성전자주식회사 , 연세대학교 산학협력단
Abstract: PURPOSE: A device and a method for recognizing feelings using voice signal are provided to extract non-linguistic parameter reflecting a emotional state from speed sound, thereby more accurately determining the feeling state of a speaker. CONSTITUTION: A non-linguistic information detector(210) detects one between pitch track and energy track from inputted speed sound as non-linguistic information. A non-linguistic parameter extractor(220) detects a specific interval from at least one between the pitch orbit and energy orbit. The non-linguistic parameter extractor extracts non-linguistic parameter from the detected specific section. A determination unit(230) determines the feeling state recognized about the speed sound through the extracted non-linguistic parameter.
Abstract translation: 目的：提供使用语音信号识别感觉的装置和方法，以从速度声音提取反映情绪状态的非语言参数，从而更准确地确定说话者的感觉状态。 构成：非语言信息检测器（210）从输入的速度声音中检测音轨和能量轨迹之间的一个作为非语言信息。 非语言参数提取器（220）从俯仰轨道和能量轨道之间的至少一个检测特定间隔。 非语言参数提取器从检测到的特定部分提取非语言参数。 确定单元（230）通过提取的非语言参数来确定关于速度声音识别的感觉状态。
-
公开(公告)号：KR1020100088461A
公开(公告)日：2010-08-09
申请号：KR1020090007706
申请日：2009-01-30
Applicant: 삼성전자주식회사 , 연세대학교 산학협력단
IPC: G10L25/63
Abstract: PURPOSE: An apparatus and a method for recognizing emotion using a voice signal are provided to accurately judge the emotional state of a speaker by using a parameter which reflects the utterance characteristics of a general parameter which reflects the emotional state. CONSTITUTION: A normal characteristic parameter extractor(210) extracts a general characteristic parameter, which indicates an emotional state, from an inputted utterance sound. An utterance characteristic parameter extractor(220) extracts the utterance characteristic parameter which indicates an inputted utterance sound. The extracted normal characteristic parameter and a decision unit(230) judges the emotional state by using the extracted utterance characteristic parameter.
Abstract translation: 目的：提供一种用于使用语音信号识别情感的装置和方法，以通过使用反映反映情绪状态的一般参数的发音特征的参数来准确地判断说话者的情绪状态。 构成：正常特征参数提取器（210）从输入的话音中提取表示情绪状态的一般特征参数。 发声特征参数提取器（220）提取表示输入的发声的发声特性参数。 所提取的正常特征参数和判定单元（230）通过使用所提取的话语特征参数来判断情绪状态。
-
公开(公告)号：KR1020090019113A
公开(公告)日：2009-02-25
申请号：KR1020070083299
申请日：2007-08-20
Applicant: 삼성전자주식회사
Inventor： 유성엽
IPC: G10L21/0208 , G10L15/20 , G10L25/00
CPC classification number: G10L21/0208 , G10L19/02 , G10L25/21
Abstract: An apparatus and a method for eradicating a noise of voice signal are provided to adjust noise eradication and voice distortion according to surrounding environment and the variation of noise characteristics when eradicating noise by using an EVRC(Enhanced Variable Rate Coder) in a mobile communication system. A noise eradication device receives a voice signal frame(300), and measures the signal energy of each channel from the received voice signal(302). The noise eradication device the noises of each channel(304), and measures an SNR(Signal to Noise Ratio) for each channel in a current frame by obtaining not only the noise energy ratio of each channel in a previous frame but the energy value of each channel(306). The noise eradication device calculates an MOG(Minimum Overall Gain) according to the SNR of each channel(308). The noise eradication device obtains a gain constant of each channel by using the MOG, and calculates the gain of each channel by using the SNR and the gain constant(310). The noise eradication device eradicates a noise by multiplying the gain of each channel to a voice spectrum of an input signal(312).
Abstract translation: 提供了一种用于消除语音信号噪声的装置和方法，以便在移动通信系统中使用EVRC（增强型可变速率编码器）来根据周围环境和消除噪声时的噪声特性的变化来调节噪声消除和语音失真。 噪声消除装置接收语音信号帧（300），并从接收到的语音信号（302）测量每个信道的信号能量。 噪声消除装置每个信道的噪声（304），并且通过不仅获得先前帧中的每个信道的噪声能量比，而且通过获得当前帧中的每个信道的能量值来测量当前帧中的每个信道的SNR（信噪比） 每个通道（306）。 噪声消除装置根据每个信道的SNR计算MOG（最小总增益）（308）。 噪声消除装置通过使用MOG获得每个信道的增益常数，并且通过使用SNR和增益常数来计算每个信道的增益（310）。 噪声消除装置通过将每个信道的增益乘以输入信号的语音频谱来消除噪声（312）。
-
Patent Agency Ranking
公开(公告)号：KR101737824B1
公开(公告)日：2017-05-19
申请号：KR1020090125690
申请日：2009-12-16
Applicant: 삼성전자주식회사
IPC: G10L21/0208 , G10L15/20
CPC classification number: H04B1/1036 , G10L21/0208 , G10L2021/02165
Abstract: 본발명의일실시예는잡음신호만으로구성된순수잡음신호또는음성신호와잡음신호가혼합된입력신호를수신하고, 그순수잡음신호가정적잡음신호인지또는비정적잡음신호인지를판단한후, 그판단결과에기초하여, 상이한잡음제거방법을이용하여입력신호로부터잡음신호를제거하는잡음제거방법을개시한다.
-
公开(公告)号：KR101560833B1
公开(公告)日：2015-10-15
申请号：KR1020090007706
申请日：2009-01-30
Applicant: 삼성전자주식회사 , 연세대학교 산학협력단
IPC: G10L25/63
Abstract: 본발명은감정을인식하는장치및 방법에관한것으로, 특히음성신호를이용하여감정을인식하는장치에있어서, 입력된발화음으로부터감정상태를나타내는일반특성파라미터를추출하는일반특성파라미터추출부; 상기입력된발화음으로부터발성특성을나타내는발성특성파라미터를추출하는발성특성파라미터추출부; 및상기추출된일반특성파라미터및 상기추출된발성특성파라미터를이용하여상기발성특성이반영된감정상태를판단하는판단부를포함한다.
-
公开(公告)号：KR1020120091701A
公开(公告)日：2012-08-20
申请号：KR1020110011624
申请日：2011-02-09
Applicant: 삼성전자주식회사 , 서울대학교산학협력단
IPC: G10L21/02
CPC classification number: G10L21/02 , G10L21/0272
Abstract: PURPOSE: A method for outputting an audio signal and a device thereof are provided to prevent sound distortion while increasing articulation of an audio signal. CONSTITUTION: A noise detecting unit(110) detects a noise signal in an inputted audio signal. The audio signal includes a target audio signal and the noise signal. A masking threshold value calculating unit(120) calculates a masking threshold value of the noise signal. An improving unit(130) amplifies the target audio signal in a frequency section during which the value of the target audio signal is higher than the masking threshold value.
Abstract translation: 目的：提供一种用于输出音频信号的方法及其装置，以防止声音失真同时增加音频信号的发音。 构成：噪声检测单元（110）检测输入的音频信号中的噪声信号。 音频信号包括目标音频信号和噪声信号。 掩蔽阈值计算单元（120）计算噪声信号的掩蔽阈值。 改善单元（130）在目标音频信号的值高于屏蔽阈值的频率部分放大目标音频信号。
-
公开(公告)号：KR1020110068637A
公开(公告)日：2011-06-22
申请号：KR1020090125690
申请日：2009-12-16
Applicant: 삼성전자주식회사
IPC: G10L21/0208 , G10L15/20
CPC classification number: H04B1/1036 , G10L21/0208 , G10L2021/02165
Abstract: PURPOSE: A method and apparatus for removing noise from an input signal are provided to reduce the distortion of a noise signal through noise estimation by the removal of the noises signal. CONSTITUTION: An apparatus receives a noise signal, a voice signal, or a mixed input signal(110). The apparatus determines whether the noise signal is a static noise signal or a non-static noise signal(120). The apparatus removes the noise signal from the input signal based on a determination result(130).
Abstract translation: 目的：提供一种从输入信号中去除噪声的方法和装置，通过去除噪声信号，通过噪声估计来减少噪声信号的失真。 构成：装置接收噪声信号，语音信号或混合输入信号（110）。 该装置确定噪声信号是静态噪声信号还是非静态噪声信号（120）。 该装置基于确定结果（130）从输入信号中去除噪声信号。
-
公开(公告)号：KR1020090032560A
公开(公告)日：2009-04-01
申请号：KR1020070097898
申请日：2007-09-28
Applicant: 삼성전자주식회사
Inventor： 유성엽
Abstract: A method for detecting a voice in an electronic device and an apparatus therefor are provided to improve the performance of detecting voice signals in various conditions by using the threshold value which is dynamically changed according to the voice signal inputted in the electronic device. A signal is inputted(301). The energy and spectrum of the inputted signal frame are measured(303). It is determined whether the inputted signal is a signal frame following the predetermined number of initial frames(305). An SNR(Signal to Noise Ratio) is measured(307). The measured SNR is compared with the preset minimum SNR value(309). The spectrum difference is calculated(315). The calculated spectrum difference is compared with the threshold value of a set corresponding frame(317). If the calculated spectrum difference is identical or smaller to or than the threshold value of the corresponding frame, the corresponding frame is determined as a silent interval(327).
Abstract translation: 提供一种用于检测电子设备中的语音的方法及其装置，用于通过使用根据在电子设备中输入的语音信号动态改变的阈值来改善在各种条件下检测语音信号的性能。 输入信号（301）。 测量输入信号帧的能量和频谱（303）。 确定输入的信号是否是预定数量的初始帧之后的信号帧（305）。 测量SNR（信噪比）（307）。 将测量的SNR与预设的最小SNR值进行比较（309）。 计算光谱差（315）。 将计算出的频谱差与设定的相应帧（317）的阈值进行比较。 如果计算出的频谱差异等于或小于对应帧的阈值，则相应的帧被确定为静默间隔（327）。
-
-
-
-
-
-
-
-
-
Search Results
10
records
(took 0.021 seconds)
