WO2018059406A1 - 盲音分离方法、结构及语音控制系统和电器总成 - Google Patents
盲音分离方法、结构及语音控制系统和电器总成 Download PDFInfo
- Publication number
- WO2018059406A1 WO2018059406A1 PCT/CN2017/103517 CN2017103517W WO2018059406A1 WO 2018059406 A1 WO2018059406 A1 WO 2018059406A1 CN 2017103517 W CN2017103517 W CN 2017103517W WO 2018059406 A1 WO2018059406 A1 WO 2018059406A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- signal
- blind
- voice
- estimating
- whitening
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0272—Voice signal separating
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0272—Voice signal separating
- G10L21/0308—Voice signal separating characterised by the type of parameter measurement, e.g. correlation techniques, zero crossing techniques or predictive techniques
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/20—Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
- G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0272—Voice signal separating
- G10L21/028—Voice signal separating using properties of sound source
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L2021/02161—Number of inputs available containing the signal or the noise to be suppressed
- G10L2021/02163—Only one microphone
Definitions
- An estimation module configured to estimate, by an iterative algorithm, a separation matrix W that is expected to be the largest of the objective function
- X(n) is a plurality of signal source speeches.
- Y X-mean(X), where X is the input data, mean(X) is the X-mean value, and Y is the data after the data centering process, and the mean value is expected to be zero.
- This step is calculated one by one, and the condition is judged in the middle to determine different starting modes of the big loop and the small loop. Specifically include:
- Step 2440 a de-correlation W p
- the blind sound separation structure can be loaded with a blind sound separation program, installed in an existing electrical assembly, or integrated into an existing voice control system in a chip manner, or loaded into an existing program hardware as a software program. in.
- a building module for constructing an objective function on the preprocessed speech signal with a non-Gaussian metric
- the preprocessing module includes a data centering unit for data centering processing, and the formula of the data centering processing is:
- An initializing unit for initial weight vector w An initializing unit for initial weight vector w
- de-correlation unit W p do decorrelation processing comprising:
- the above voice control system is applied to an electric appliance assembly, which includes an electric appliance body and the above voice control system, and the electric appliance body and the voice control system are connected.
Abstract
Description
Claims (15)
- 一种盲音分离方法,其特征在于,包括:预处理步骤,对检测的语音信号进行降噪预处理,所述语音信号为同时间多个信号源语音信息的线性叠加信号;构建步骤,以非高斯度量,对预处理后的语音信号构建目标函数;估计步骤,通过迭代算法估计所述目标函数期望最大的分离矩阵W;求取步骤,利用U(n)=WX(n)求取估计目标分离信号U(n),其中,X(n)为多个信号源语音信息构建的向量。
- 如权利要求1所述的盲音分离方法,其特征在于,所述预处理步骤包括:数据中心化处理,所述数据中心化处理的公式为:Y=X-mean(X),X为输入数据,mean(X)为X均值,Y是数据中心化处理后的数据,其均值的期望为0。
- 如权利要求1所述的盲音分离方法,其特征在于,所述预处理步骤包括:白化处理,所述白化处理的公式为:Z(t)=W0X(t),其中,W0为白化矩阵,Z为白化向量。
- 如权利要求1-4任一项所述的盲音分离方法,其特征在于,所述估计步骤包括:预估步骤,预估待估计的分量的个数m;初始化步骤,初始权矢量w;计算步骤,利用Wp=E{Zg(Wp TZ)}-E{g'(Wp TZ)}计算Wp;判断步骤,利用Wp=Wp/||Wp||,判断Wp是否收敛;若收敛,对Wp做去相关处理;令p=p+1,将p与m比较,当p<m,则返回所述初始化步骤,当p≥m时,结束,得到所述分离矩阵W。
- 如权利要求5所述的盲音分离方法,其特征在于,所述对Wp做去相关处理包括:在估计出p个向量W1,W2,,W3,W4,……Wp之后,当估计Wp+1时,先减去p个向量的投影Wp+1 TWj,j=1,…,p,然后标准化Wp+1。
- 如权利要求5所述盲音分离方法,其特征在于,若不收敛,则返回所述计算步骤。
- 一种执行权利要求1-7任一项所述的盲音分离方法的盲音分离结构,其特征在于,包括:预处理模块,用于对检测的语音信号进行降噪预处理,所述语音信号为同时间多个信号源语音信息的线性叠加信号;构建模块,用于以非高斯度量,对预处理后的语音信号构建目标函数;估计模块,用于通过迭代算法估计所述目标函数期望最大的分离矩阵W;求取模块,用于利用U(n)=WX(n)求取估计目标分离信号U(n),其中,X(n)为多个信号源语音信息构建的向量。
- 如权利要求8所述的盲音分离结构,其特征在于,所述预处理模块包括数据中心化单元,其用于数据中心化处理,所述数据中心化处理的公式为:Y=X-mean(X),X为输入数据,mean(X)为X均值,Y是数据中心化处理后的数据,其均值的期望为0。
- 如权利要求8所述的盲音分离结构,其特征在于,所述预处理模块包括白化单元,其用于白化处理,所述白化处理的公式为:Z(t)=W0X(t),其中,W0为白化矩阵,Z为白化向量。
- 如权利要求8-10任一项所述的盲音分离结构,其特征在于,所述估计模块包括:预估单元,用于预估待估计的分量的个数m;初始化单元,用于初始权矢量w;计算单元,用于利用Wp=E{Zg(Wp TZ)}-E{g'(Wp TZ)}计算Wp;判断单元,利用Wp=Wp/||Wp||,判断Wp是否收敛,若收敛,激活去相关单元对Wp做去相关处理,令p=p+1;将p与m比较,当p<m,则返 回所述初始化单元,当p≥m时,结束,得到所述分离矩阵W;若不收敛,激活所述计算单元。
- 如权利要求11所述的盲音分离结构,其特征在于,所述去相关单元对Wp做去相关处理包括:在估计出p个向量W1,W2,,W3,W4,……Wp之后,当估计Wp+1时,先减去p个向量的投影Wp+1 TWj,j=1,…,p,然后标准化Wp+1。
- 一种语音控制系统,其特征在于,包括:语音检测组件和权利要求8-12任一项所述的盲音分离结构;所述语音检测组件,用于检测环境内多个信号源语音信息,得到语音信号,供所述盲音分离结构进行盲音分离。
- 一种电器总成,其特征在于,包括电器本体和权利要求13所述的语音控制系统,所述电器本体和所述语音控制系统相连。
- 如权利要求14所述的电器总成,其特征在于,包括:家用电器、中央空调、电子类移动终端中的任一种。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US16/338,198 US10825466B2 (en) | 2016-09-29 | 2017-09-26 | Blind signal separation method and structure, voice control system, and electrical appliance assembly |
EP17854856.6A EP3522156A4 (en) | 2016-09-29 | 2017-09-26 | BLINDER SIGNAL SEPARATION AND STRUCTURE, LANGUAGE CONTROL SYSTEM AND ELECTRICAL EQUIPMENT ASSEMBLY |
JP2019517820A JP6790253B2 (ja) | 2016-09-29 | 2017-09-26 | ブラインド信号分離方法、構成及び音声制御システム、並びに電器アセンブリ |
KR1020197012155A KR20190054157A (ko) | 2016-09-29 | 2017-09-26 | 블라인드 신호 분리 방법과 구조, 음성 제어 시스템 및 전기 장치 어셈블리 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610866508.9A CN106356075B (zh) | 2016-09-29 | 2016-09-29 | 盲音分离方法、结构及语音控制系统和电器总成 |
CN201610866508.9 | 2016-09-29 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2018059406A1 true WO2018059406A1 (zh) | 2018-04-05 |
Family
ID=57867241
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2017/103517 WO2018059406A1 (zh) | 2016-09-29 | 2017-09-26 | 盲音分离方法、结构及语音控制系统和电器总成 |
Country Status (6)
Country | Link |
---|---|
US (1) | US10825466B2 (zh) |
EP (1) | EP3522156A4 (zh) |
JP (1) | JP6790253B2 (zh) |
KR (1) | KR20190054157A (zh) |
CN (1) | CN106356075B (zh) |
WO (1) | WO2018059406A1 (zh) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112116922A (zh) * | 2020-09-17 | 2020-12-22 | 集美大学 | 一种噪声盲源信号分离方法、终端设备及存储介质 |
Families Citing this family (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106356075B (zh) * | 2016-09-29 | 2019-09-17 | 合肥美的智能科技有限公司 | 盲音分离方法、结构及语音控制系统和电器总成 |
CN109413543B (zh) * | 2017-08-15 | 2021-01-19 | 音科有限公司 | 一种源信号提取方法、系统和存储介质 |
CN109994120A (zh) * | 2017-12-29 | 2019-07-09 | 福州瑞芯微电子股份有限公司 | 基于双麦的语音增强方法、系统、音箱及存储介质 |
WO2020172831A1 (en) * | 2019-02-28 | 2020-09-03 | Beijing Didi Infinity Technology And Development Co., Ltd. | Concurrent multi-path processing of audio signals for automatic speech recognition systems |
WO2021100136A1 (ja) * | 2019-11-20 | 2021-05-27 | 日本電信電話株式会社 | 音源信号推定装置、音源信号推定方法、プログラム |
CN111312276B (zh) * | 2020-02-14 | 2023-01-17 | 北京声智科技有限公司 | 一种音频信号处理的方法、装置、设备和介质 |
CN113674752B (zh) * | 2020-04-30 | 2023-06-06 | 抖音视界有限公司 | 音频信号的降噪方法、装置、可读介质和电子设备 |
CN111863020B (zh) * | 2020-07-30 | 2022-09-20 | 腾讯科技(深圳)有限公司 | 语音信号处理方法、装置、设备及存储介质 |
CN112082793A (zh) * | 2020-08-31 | 2020-12-15 | 洛阳师范学院 | 一种基于SCA和FastICA的旋转机械耦合故障诊断方法 |
CN113470689B (zh) * | 2021-08-23 | 2024-01-30 | 杭州国芯科技股份有限公司 | 一种语音分离方法 |
CN113794489B (zh) * | 2021-09-07 | 2022-12-20 | 中国人民解放军陆军工程大学 | 一种通信抗强相关干扰的方法 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070133811A1 (en) * | 2005-12-08 | 2007-06-14 | Kabushiki Kaisha Kobe Seiko Sho | Sound source separation apparatus and sound source separation method |
CN101833955A (zh) * | 2010-01-22 | 2010-09-15 | 大连理工大学 | 一种基于负熵最大化的复数约束独立分量分析方法 |
CN104064186A (zh) * | 2014-06-26 | 2014-09-24 | 山东大学 | 一种基于独立分量分析的电气设备故障音检测方法 |
CN106356075A (zh) * | 2016-09-29 | 2017-01-25 | 合肥华凌股份有限公司 | 盲音分离方法、结构及语音控制系统和电器总成 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB0204548D0 (en) * | 2002-02-27 | 2002-04-10 | Qinetiq Ltd | Blind signal separation |
US8874439B2 (en) * | 2006-03-01 | 2014-10-28 | The Regents Of The University Of California | Systems and methods for blind source signal separation |
JP5078717B2 (ja) * | 2008-04-08 | 2012-11-21 | 三菱電機株式会社 | 入射波数推定装置及び入射波数推定方法 |
JP2011107603A (ja) * | 2009-11-20 | 2011-06-02 | Sony Corp | 音声認識装置、および音声認識方法、並びにプログラム |
JP5408810B2 (ja) * | 2011-06-24 | 2014-02-05 | アイシン・エィ・ダブリュ株式会社 | 音声認識制御システム、音声認識制御方法、及び音声認識制御プログラム |
-
2016
- 2016-09-29 CN CN201610866508.9A patent/CN106356075B/zh active Active
-
2017
- 2017-09-26 WO PCT/CN2017/103517 patent/WO2018059406A1/zh unknown
- 2017-09-26 US US16/338,198 patent/US10825466B2/en active Active
- 2017-09-26 EP EP17854856.6A patent/EP3522156A4/en not_active Withdrawn
- 2017-09-26 KR KR1020197012155A patent/KR20190054157A/ko not_active IP Right Cessation
- 2017-09-26 JP JP2019517820A patent/JP6790253B2/ja active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070133811A1 (en) * | 2005-12-08 | 2007-06-14 | Kabushiki Kaisha Kobe Seiko Sho | Sound source separation apparatus and sound source separation method |
CN101833955A (zh) * | 2010-01-22 | 2010-09-15 | 大连理工大学 | 一种基于负熵最大化的复数约束独立分量分析方法 |
CN104064186A (zh) * | 2014-06-26 | 2014-09-24 | 山东大学 | 一种基于独立分量分析的电气设备故障音检测方法 |
CN106356075A (zh) * | 2016-09-29 | 2017-01-25 | 合肥华凌股份有限公司 | 盲音分离方法、结构及语音控制系统和电器总成 |
Non-Patent Citations (3)
Title |
---|
CHEN, YAN: "Application of Improved independent Component Analysis Technology in Speech Signal Separation", ELECTRONIC SCIENCE AND TECHNOLOGY, vol. 22, no. 12, 15 December 2009 (2009-12-15), pages 83 - 87, XP009513559, ISSN: 1007-7820 * |
QIU, ZUOCHUN: "Application of ICA in signal separation and denoising", POPULAR SCIENCE & TECHNOLOGY, no. 12, 31 December 2009 (2009-12-31), pages 28 - 29, XP009518731, ISSN: 1008-1151 * |
See also references of EP3522156A4 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112116922A (zh) * | 2020-09-17 | 2020-12-22 | 集美大学 | 一种噪声盲源信号分离方法、终端设备及存储介质 |
CN112116922B (zh) * | 2020-09-17 | 2024-04-12 | 集美大学 | 一种噪声盲源信号分离方法、终端设备及存储介质 |
Also Published As
Publication number | Publication date |
---|---|
US20200027473A1 (en) | 2020-01-23 |
EP3522156A1 (en) | 2019-08-07 |
EP3522156A4 (en) | 2019-10-30 |
JP2019533194A (ja) | 2019-11-14 |
CN106356075A (zh) | 2017-01-25 |
US10825466B2 (en) | 2020-11-03 |
JP6790253B2 (ja) | 2020-11-25 |
CN106356075B (zh) | 2019-09-17 |
KR20190054157A (ko) | 2019-05-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2018059406A1 (zh) | 盲音分离方法、结构及语音控制系统和电器总成 | |
JP7186769B2 (ja) | 対象音声検出方法及び装置 | |
JP2019533193A (ja) | 音声制御システム及びそのウェイクアップ方法、ウェイクアップ装置、並びに家電製品、コプロセッサ | |
CN105704298A (zh) | 声音唤醒侦测装置与方法 | |
Liao et al. | Power disaggregation for low-sampling rate data | |
Kobayashi et al. | Acoustic feature extraction by statistics based local binary pattern for environmental sound classification | |
WO2021008000A1 (zh) | 语音唤醒方法、装置及电子设备、存储介质 | |
Zhou et al. | Mixing matrix estimation from sparse mixtures with unknown number of sources | |
WO2013089536A1 (ko) | 관심음원 제거방법 및 그에 따른 음성인식방법 및 음성인식장치 | |
CN113591736A (zh) | 特征提取网络、活体检测模型的训练方法和活体检测方法 | |
WO2023273747A1 (zh) | 智能设备的唤醒方法和装置、存储介质及电子装置 | |
Srinivasarengan et al. | A framework for non intrusive load monitoring using bayesian inference | |
Kumar et al. | Data-driven electrical load disaggregation using graph signal processing | |
Kumar et al. | An intuitive explanation of graph signal processing-based electrical load disaggregation | |
CN110190832B (zh) | 变正则化参数多任务自适应滤波器网络 | |
KR101711302B1 (ko) | 변별적 가중치 학습기법을 이용한 2 채널 마이크 기반의 음성 검출 장치 및 그 방법 | |
Azizan et al. | A stochastic interpretation of stochastic mirror descent: Risk-sensitive optimality | |
Shin et al. | An affine projection algorithm with update-interval selection | |
WO2019183747A1 (zh) | 语音检测的方法和装置 | |
Djurovic et al. | QQ-plot based probability density function estimation | |
CN112185367A (zh) | 关键词检测方法和装置、计算机可读存储介质、电子设备 | |
Ito | Recognition of sounds using square cauchy mixture distribution | |
Chen et al. | Acoustic vector sensor based speech source separation with mixed Gaussian-Laplacian distributions | |
Srikrishnan et al. | Improved noise characterization for Relative Impulse Response Estimation | |
Liu et al. | An adaptive non local spatial fuzzy image segmentation algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17854856 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2019517820 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 20197012155 Country of ref document: KR Kind code of ref document: A |
|
ENP | Entry into the national phase |
Ref document number: 2017854856 Country of ref document: EP Effective date: 20190429 |