WO2016056683A1 - 전자 장치 및 이의 잔향 제거 방법 - Google Patents
전자 장치 및 이의 잔향 제거 방법 Download PDFInfo
- Publication number
- WO2016056683A1 WO2016056683A1 PCT/KR2014/009405 KR2014009405W WO2016056683A1 WO 2016056683 A1 WO2016056683 A1 WO 2016056683A1 KR 2014009405 W KR2014009405 W KR 2014009405W WO 2016056683 A1 WO2016056683 A1 WO 2016056683A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- reverberation
- user
- component
- voice
- initial
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012805 post-processing Methods 0.000 claims abstract description 13
- 230000000903 blocking effect Effects 0.000 claims description 29
- 239000011159 matrix material Substances 0.000 claims description 28
- 230000003044 adaptive effect Effects 0.000 claims description 9
- 238000000926 separation method Methods 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 3
- 210000004392 genitalia Anatomy 0.000 claims description 2
- 210000003899 penis Anatomy 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 239000000284 extract Substances 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 230000014509 gene expression Effects 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
- G10L21/0232—Processing in the frequency domain
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M1/00—Substation equipment, e.g. for use by subscribers
- H04M1/60—Substation equipment, e.g. for use by subscribers including speech amplifiers
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R1/00—Details of transducers, loudspeakers or microphones
- H04R1/20—Arrangements for obtaining desired frequency or directional characteristics
- H04R1/32—Arrangements for obtaining desired frequency or directional characteristics for obtaining desired directional characteristic only
- H04R1/40—Arrangements for obtaining desired frequency or directional characteristics for obtaining desired directional characteristic only by combining a number of identical transducers
- H04R1/406—Arrangements for obtaining desired frequency or directional characteristics for obtaining desired directional characteristic only by combining a number of identical transducers microphones
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R3/00—Circuits for transducers, loudspeakers or microphones
- H04R3/005—Circuits for transducers, loudspeakers or microphones for combining the signals of two or more microphones
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R3/00—Circuits for transducers, loudspeakers or microphones
- H04R3/02—Circuits for transducers, loudspeakers or microphones for preventing acoustic reaction, i.e. acoustic oscillatory feedback
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
- G10L2021/02082—Noise filtering the noise being echo, reverberation of the speech
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
- G10L2021/02166—Microphone arrays; Beamforming
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2430/00—Signal processing covered by H04R, not provided for in its groups
- H04R2430/20—Processing of the output signals of the acoustic transducers of an array for obtaining a desired directivity characteristic
- H04R2430/25—Array processing for suppression of unwanted side-lobes in directivity characteristics, e.g. a blocking matrix
Definitions
- the present invention relates to an electronic device and a method for removing reverberation thereof, and more particularly, to an electronic device and a method for removing reverberation thereof that can obtain an initial component by removing a reverberation component from a user voice obtained from a plurality of microphones.
- ASR automatic speech recognition
- a technique of removing the reverberation component using one microphone is provided, but a technique of removing the reverberation component using one microphone is excellent in certain conditions.
- a technique of removing the reverberation component using one microphone does not effectively remove the reverberation component.
- the present invention has been made to solve the above-described problem, and an object of the present invention is to provide an electronic device capable of removing reverberation components from a user's voice obtained using a plurality of microphones, and a method for removing reverberation thereof.
- An electronic device for achieving the above object, a plurality of microphone unit for receiving a user's voice; A reverberation remover configured to remove reverberation components of the user voice received from the plurality of microphone units to obtain initial components of the user voice; Reverberation information obtaining unit obtaining intensity information on the reverberation component of the user's voice; And a post-processing unit which further removes the reverberation component from the initial component obtained from the reverberation removing unit based on the intensity information on the reverberation component.
- the reverberation remover may acquire an initial component of the user voice by removing a reverberation component of the user voice using a minimum variance distortionless response (MVDR) beamformer.
- MVDR minimum variance distortionless response
- the MVDR beamformer may be implemented as a generalized sidelobe canceller.
- the general purpose side lobe remover may include: a fixed beamformer for extracting user voices of initial components among user voices received from the plurality of microphone units; A blocking matrix to block initial components of user voices received from the plurality of microphones; And an adaptive noise canceller for adaptively removing reverberation components from the user voice of the initial component using the user voice from which the genital initial component has been removed.
- the fixed beamformer may be a delay beam (DS) beamformer.
- DS delay beam
- an initial component measuring unit measuring an initial component from the plurality of microphone units, wherein the blocking matrix is a user received from the plurality of microphone units using the initial component output from the initial component measuring unit. Block initial components of the negative.
- the reverberation canceller may include: a fixed beamformer configured to extract a user voice of an initial component among user voices received from the plurality of microphone units; And a block matrix based on a blind source separation (BSS) that blocks initial components of user voices received from the plurality of microphones.
- BSS blind source separation
- the post processor may remove the reverberation component of the user voice by filtering the user voice from which the initial component output from the BSS-based blocking matrix is removed from the user voice of the initial component output from the fixed beamformer. .
- a method for removing reverberation of an electronic device comprising: receiving a user's voice through a plurality of microphones; Obtaining an initial component of the user's voice by removing a reverberation component of the user's voice received from the plurality of microphones; Obtaining intensity information on the reverberation component of the user voice; And a post-processing step of further removing the reverberation component from the obtained initial component based on the intensity information on the reverberation component.
- the acquiring of an initial component of the user voice may be performed by removing a reverberation component of the user voice by using a minimum variance distortionless response (MVDR) beamformer.
- MVDR minimum variance distortionless response
- the MVDR beamformer may be implemented as a generalized sidelobe canceller.
- the acquiring of an initial component of the user voice may include extracting a user voice of the initial component among the user voices received from the plurality of microphones using a fixed beamformer; Blocking an initial component of user voices received from the plurality of microphones by using a blocking matrix; And adaptively removing the reverberation component from the user voice of the initial component by using the user voice from which the initial component of the penis has been removed by using an adaptive noise canceller.
- the fixed beamformer may be a delay beam (DS) beamformer.
- DS delay beam
- the method may further include measuring initial components from the plurality of microphone units, wherein the blocking matrix may block initial components of user voices received from the plurality of microphone units using the initial components. Can be.
- the acquiring of an initial component of the user voice may include extracting a user voice of the initial component from among the user voices received from the plurality of microphone units using a fixed beamformer; And blocking an initial component of user voices received from the plurality of microphones using a blind source separation (BSS) based blocking matrix.
- BSS blind source separation
- the post-processing step may remove the reverberation component of the user voice by filtering the user voice from which the initial component output from the BSS-based blocking matrix is removed from the user voice of the initial component output from the fixed beamformer.
- various embodiments of the present invention enable the electronic device to improve the voice recognition rate in an environment with high noise and reverberation.
- FIG. 1 is a block diagram illustrating a configuration of an electronic device for removing a reverberation component according to an embodiment of the present disclosure
- FIG. 2 is a block diagram illustrating in detail a configuration of an electronic device for removing a reverberation component according to an embodiment of the present disclosure
- FIG. 3 is a block diagram illustrating in detail a configuration of an electronic device for removing a reverberation component according to another embodiment of the present disclosure
- FIG. 4 is a flowchart illustrating a method of removing reverberation of an electronic device according to an embodiment of the present disclosure.
- first and second may be used to describe various components, but the components should not be limited by the terms. The terms are only used to distinguish one component from another.
- the module or unit performs at least one function or operation, and may be implemented by hardware or software, or a combination of hardware and software.
- the plurality of modules or the plurality of units may be integrated into at least one module except for the modules or units that need to be implemented with specific hardware, and are implemented as at least one processor (not shown). Can be.
- the electronic device 100 includes a plurality of microphone units 110, reverberation removing unit 120, reverberation information obtaining unit 130, and post-processing unit 140.
- the electronic device 100 may be a device including a function of recognizing a user voice, such as a smart phone, a smart TV, a tablet PC, a notebook PC, a desktop PC, and the like.
- the plurality of microphones 110 receive a user's voice.
- the plurality of micro parts 110 may be arranged in a single row.
- the reverberation remover 120 removes reverberation components of the user voices received from the plurality of microphones 110 to obtain initial components of the user voices.
- the reverberation remover 120 may obtain an initial component of the user voice by removing the reverberation component of the user voice using a beamforming technique.
- beamforming is an audio processing technique for improving directivity by removing energy received from other directions while maintaining user voice received from a target direction.
- beamformers are generally used to remove noise and interference.
- the beamformer can reduce the level of the reverberant component incident in the other direction, thus providing a direct-to-reverberant ratio (DRR).
- DRR direct-to-reverberant ratio
- the reverberation remover 120 may remove the reverberation component of the user's voice by using the MVDR beamformer.
- MVDR beamformers are isotropic and allow the highest directivity in the diffused sound field.
- the MVDR beamformer may be implemented with a Generalized Sidelobe Canceller (GSC). A method of removing the reverberation component of the user voice by using the general purpose side lobe remover will be described with reference to FIG. 2.
- the reverberation component of the user's voice may be removed using the MVDR beamformer, but this is only an example, and the reverberation component of the user's voice may be removed using another type of beamformer.
- the reverberation remover 120 may remove the reverberation component of the user voice by using a blind source separation (BSS) based blocking matrix.
- BSS blind source separation
- the reverberation information obtaining unit 130 obtains intensity information on the reverberation component of the user voice input from the plurality of microphone units 110.
- the reverberation information obtaining unit 130 may obtain a power spectral density (PSD) for the reverberation component of the user's voice.
- PSD power spectral density
- the post processor 140 may additionally remove the reverberation component from the initial component obtained from the reverberation remover 120 based on the intensity information on the reverberation component obtained from the reverberation information obtaining unit 130.
- the electronic device 200 includes a plurality of microphones 210, a GSC 220, an initial component measurer 230, a reverberation information acquirer 240, and a post processor 250. Include.
- the plurality of microphones 210 receive a user voice.
- the plurality of microphones 210 may be arranged in one row.
- each of the microphone units 210 may remove a part of the reverberation component using a single channel dereverberation algorithm.
- the GSC unit 220 may obtain an initial component of the user's voice by removing the reverberation component of the user's voice received from the plurality of microphones 210.
- the GSC unit 220 includes a fixed beamformer 221, a blocking matrix 223, an adaptive noise canceller 225, and an adder 227, as shown in FIG. 2.
- the fixed beamformer 221 extracts user voices of initial components among user voices received from the plurality of microphones 210.
- the fixed beamformer 221 may be implemented with a delay and sum (DS) beamformer to reduce the level of the reverberation component.
- DS delay and sum
- the fixed beamformer 221 receives signals of different phases through the plurality of microphones 210 to compensate only phases of the target signal, and then adds signals of each channel to reduce the reverberation signal of the user. Voice can be extracted. In this case, since the fixed beamformer 221 compensates only the phase of the initial component of the user voice using N microphones, the size of the reverberation signal is reduced to 1 / N.
- the blocking matrix 223 removes initial components of user voices received from the plurality of microphones 210.
- the blocking matrix 223 may remove initial components of user voices received from the plurality of microphone units by using the initial components measured by the initial component measurer 230.
- the reverberation component of the user voice may be removed from all the microphones 210.
- the initial component measurer 230 measures an initial component of the user voice signal from which a part of the reverberation component is removed from the plurality of microphones 210.
- the initial component of the user's voice is used to measure the relative transfer function (RTF) associated with the initial component of the user's voice by a least-squares procedure.
- the blocking matrix 223 may remove the initial component from the user voice and generate a reference signal for the reverberation component by using an RTF associated with the initial component of the user voice.
- the adaptive noise canceller 225 adaptively removes the reverberation component of the user voice of the initial component by using the user voice from which the initial component has been removed. Specifically, the adaptive noise canceller 225 uses the reference signal for the reverberation component that is the output of the blocking matrix 223 to add the residual reverberation component of the user voice to minimize the reverberation component at the output of the fixed beamformer 221. It can be calculated adaptively.
- the reverberation component may be removed from the initial component of the user's voice output from the fixed beamformer 221 by the adder 227.
- the adaptive noise canceller 225 has a closed-form multichannel Wiener filter because the reverberation component is non-stationary and the voice signal changes with time. ) Can be implemented.
- the adaptive noise canceller 225 may adaptively calculate the reverberation component of the user's voice using the intensity information of the reverberation component obtained from the reverberation information obtaining unit 240.
- the reverberation information obtaining unit 240 obtains intensity information on the reverberation component among the user voices obtained from the plurality of microphone units 210.
- the reverberation information acquisition unit 240 may acquire a power spectral density (PSD) for the reverberation component of the user's voice.
- PSD power spectral density
- the post processor 250 further removes the reverberation component from the user voice of the initial component from which the reverberation component is removed from the GSC unit 220 using the intensity information on the reverberation component obtained from the reverberation information obtaining unit 240.
- the post-processing unit 250 uses the coherent matrix of the reverberation and noise to form the GSC unit.
- the residual reverberation component of the initial component output from 220 may be calculated and the residual reverberation component may be further removed by using a single-channel postfilter.
- the post processor 250 may remove the reverberation component of the room impulse response based on the intensity information on the reverberation component obtained from the reverberation information acquisition unit 240.
- the post-processing unit 250 may output an initial component of the user's voice from which the residual reverberation component has been removed to the speech recognition unit (not shown).
- the electronic device 200 may improve the success rate of the voice recognition.
- FIG. 3 is a block diagram illustrating a detailed configuration of an electronic device for removing reverberation component using blind source separation (BSS) technology according to another embodiment of the present invention.
- BSS blind source separation
- the electronic device 300 includes a plurality of microphones 310, a BSS-based reverberation remover 320, an initial component measurer 330, a reverberation information acquirer 340, and a post processor. And 350.
- the plurality of microphone units 310 the initial component measuring unit 330, the reverberation information obtaining unit 340, and the post-processing unit 350, the plurality of microphone units 310 and the initial component measuring unit ( The description overlapping with the reverberation information obtaining unit 340 and the post processing unit 350 will be omitted.
- the BSS-based reverberation canceller 320 includes a fixed beamformer 321 and a BSS-based blocking matrix 323.
- the fixed beamformer 321 extracts user voices of initial components among user voices received from the plurality of microphone units 310, like the fixed beamformer 221 described with reference to FIG. 2.
- the fixed beamformer 321 may be implemented as a delay and sum (DS) beamformer to reduce the level of the reverberation component.
- the BSS-based blocking matrix 323 separates the initial component of the user voice and the reverberation script based on the information on the initial component of the user voice obtained from the initial component acquirer 330, and processes the separated reverberation component into a post-processing unit. Output to 250.
- the BSS-based blocking matrix 323 may extract the reverberation component assuming that the reverberation component is located in a null in the initial component, conceptually.
- the BSS-based blocking matrix 323 may design a cost function that includes two constraints to separate the speech signal from the mixed signal and remove the initial component.
- the two constraints include 1) decorrelation of the BSS algorithm output and 2) reverberation constraint.
- the post-processing unit 350 may perform the BSS-based blocking matrix 323 on the initial component of the user voice output from the fixed beamformer 321 based on the intensity information of the reverberation component obtained from the reverberation information obtaining unit 340.
- the output reverberation component can be removed.
- the electronic device 200 may improve the success rate of the voice recognition.
- FIG. 4 is a flowchart illustrating a method of removing reverberation of an electronic device according to an embodiment of the present disclosure.
- the electronic device 100 receives a user's voice through a plurality of microphones (S410).
- the electronic device 100 obtains an initial component of the user's voice by removing the reverberation component of the user's voice received from the plurality of microphones.
- the electronic device 100 may remove the reverberation component using GSC, or as described with reference to FIG. 3, by using the BSS technique.
- the electronic device 100 obtains intensity information on the reverberation component of the user's voice.
- the electronic device 100 may obtain information about the PSD of the reverberation component.
- the electronic device 100 may further remove the reverberation component from the initial component obtained based on the intensity information on the reverberation component.
- the electronic device 100 may acquire a more accurate initial component of the user's voice, thereby providing a high performance voice recognition function.
- connection or connection members of the lines between the components shown in the drawings by way of example shows a functional connection and / or physical or circuit connections, in the actual device replaceable or additional various functional connections, physical It may be represented as a connection, or circuit connections.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Acoustics & Sound (AREA)
- Physics & Mathematics (AREA)
- Otolaryngology (AREA)
- Human Computer Interaction (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Quality & Reliability (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Circuit For Audible Band Transducer (AREA)
Abstract
Description
Claims (15)
- 사용자의 음성을 입력받는 복수의 마이크부;상기 복수의 마이크부로부터 수신된 사용자 음성의 잔향 성분을 제거하여 사용자 음성의 초기 성분을 획득하는 잔향 제거부;상기 사용자 음성의 잔향 성분에 대한 세기 정보를 획득하는 잔향 정보 획득부;상기 잔향 성분에 대한 세기 정보를 바탕으로 상기 잔향 제거부로부터 획득된 초기 성분에 잔향 성분을 추가로 제거하는 후처리부;를 포함하는 전자 장치.
- 제1항에 있어서,상기 잔향 제거부는,MVDR(Minimum variance Distortionless Response) 빔포머(Beamformer)를 이용하여 상기 사용자 음성의 잔향 성분을 제거하여 사용자 음성의 초기 성분을 획득하는 것을 특징으로 하는 전자 장치.
- 제2항에 있어서,상기 MVDR 빔포머는,범용 사이드로브 제거기(Generalized Sidelobe Canceller)로 구현되는 것을 특징으로 하는 전자 장치.
- 제3항에 있어서,상기 범용 사이드로브 제거기는,상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분의 사용자 음성을 추출하는 고정형 빔포머(fixed beamformer);상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분을 제거(block)하는 차단 매트릭스(blocking matrix); 및성기 초기 성분이 제거된 사용자 음성을 이용하여 초기 성분의 사용자 음성 중 잔향 성분을 적응적으로 제거하는 적응적 노이즈 제거기(adaptive noise canceller);를 포함하는 전자 장치.
- 제4항에 있어서,상기 고정형 빔포머는,DS(Delay and Sum) 빔포머인 것을 특징으로 하는 전자 장치.
- 제4항에 있어서,상기 복수의 마이크부로부터 초기 성분을 측정하는 초기 성분 측정부;를 더 포함하고,상기 차단 매트릭스는,상기 초기 성분 측정부로부터 출력되는 초기 성분을 이용하여 상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분을 제거(block)하는 것을 특징으로 하는 전자 장치.
- 제1항에 있어서,상기 잔향 제거부는,상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분의 사용자 음성을 추출하는 고정형 빔포머(fixed beamformer); 및상기 복수의 마이크로부터 수신된 사용자 음성 중 초기 성분을 제거(block)하는 BSS(Blind Source Separation) 기반의 차단 매트릭스;를 포함하는 것을 특징으로 하는 전자 장치.
- 제7항에 있어서,상기 후처리부는,상기 고정형 빔포머로부터 출력되는 초기 성분의 사용자 음성에 상기 BSS 기반의 차단 매트릭스로부터 출력된 초기 성분이 제거된 사용자 음성을 필터링하여 상기 사용자 음성의 잔향 성분을 제거하는 것을 특징으로 하는 전자 장치.
- 전자 장치의 잔향 제거 방법에 있어서,복수의 마이크를 통해 사용자의 음성을 입력받는 단계;상기 복수의 마이크로부터 수신된 사용자 음성의 잔향 성분을 제거하여 사용자 음성의 초기 성분을 획득하는 단계;상기 사용자 음성의 잔향 성분에 대한 세기 정보를 획득하는 단계;상기 잔향 성분에 대한 세기 정보를 바탕으로 상기 획득된 초기 성분에 잔향 성분을 추가로 제거하는 후처리 단계;를 포함하는 잔향 제거 방법.
- 제9항에 있어서,상기 사용자 음성의 초기 성분을 획득하는 단계는,MVDR(Minimum variance Distortionless Response) 빔포머(Beamformer)를 이용하여 상기 사용자 음성의 잔향 성분을 제거하여 사용자 음성의 초기 성분을 획득하는 것을 특징으로 하는 잔향 제거 방법.
- 제9항에 있어서,상기 MVDR 빔포머는,범용 사이드로브 제거기(Generalized Sidelobe Canceller)로 구현되는 것을 특징으로 하는 잔향 제거 방법.
- 제11항에 있어서,상기 사용자 음성의 초기 성분을 획득하는 단계는,고정형 빔포머(fixed beamformer)를 이용하여 상기 복수의 마이크로부터 수신된 사용자 음성 중 초기 성분의 사용자 음성을 추출하는 단계;차단 매트릭스(blocking matrix)를 이용하여 상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분을 제거(block)하는 ; 및적응적 노이즈 제거기(adaptive noise canceller)를 이용하여성기 초기 성분이 제거된 사용자 음성을 이용하여 초기 성분의 사용자 음성 중 잔향 성분을 적응적으로 제거하는 단계;를 포함하는 잔향 제거 방법.
- 제12항에 있어서,상기 고정형 빔포머는,DS(Delay and Sum) 빔포머인 것을 특징으로 하는 잔향 제거 방법.
- 제12항에 있어서,상기 복수의 마이크부로부터 초기 성분을 측정하는 단계;를 더 포함하고,상기 차단 매트릭스는,상기 초기 성분을 이용하여 상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분을 제거(block)하는 것을 특징으로 하는 잔향 제거 방법.
- 제9항에 있어서,상기 사용자 음성의 초기 성분을 획득하는 단계는,고정형 빔포머(fixed beamformer)를 이용하여 상기 복수의 마이크부로부터 수신된 사용자 음성 중 초기 성분의 사용자 음성을 추출하는 단계; 및BSS(Blind Source Separation) 기반의 차단 매트릭스를 이용하여 상기 복수의 마이크로부터 수신된 사용자 음성 중 초기 성분을 제거(block)하는 단계;를 포함하는 것을 특징으로 하는 잔향 제거 방법.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020177008048A KR20170063618A (ko) | 2014-10-07 | 2014-10-07 | 전자 장치 및 이의 잔향 제거 방법 |
PCT/KR2014/009405 WO2016056683A1 (ko) | 2014-10-07 | 2014-10-07 | 전자 장치 및 이의 잔향 제거 방법 |
US15/517,855 US9997170B2 (en) | 2014-10-07 | 2014-10-07 | Electronic device and reverberation removal method therefor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/KR2014/009405 WO2016056683A1 (ko) | 2014-10-07 | 2014-10-07 | 전자 장치 및 이의 잔향 제거 방법 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2016056683A1 true WO2016056683A1 (ko) | 2016-04-14 |
Family
ID=55653285
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2014/009405 WO2016056683A1 (ko) | 2014-10-07 | 2014-10-07 | 전자 장치 및 이의 잔향 제거 방법 |
Country Status (3)
Country | Link |
---|---|
US (1) | US9997170B2 (ko) |
KR (1) | KR20170063618A (ko) |
WO (1) | WO2016056683A1 (ko) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017218128A1 (en) * | 2016-06-15 | 2017-12-21 | Intel Corporation | Far field automatic speech recognition pre-processing |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10403299B2 (en) * | 2017-06-02 | 2019-09-03 | Apple Inc. | Multi-channel speech signal enhancement for robust voice trigger detection and automatic speech recognition |
US10418048B1 (en) * | 2018-04-30 | 2019-09-17 | Cirrus Logic, Inc. | Noise reference estimation for noise reduction |
CN110428852B (zh) * | 2019-08-09 | 2021-07-16 | 南京人工智能高等研究院有限公司 | 语音分离方法、装置、介质和设备 |
DK3863303T3 (da) | 2020-02-06 | 2023-01-16 | Univ Zuerich | Vurdering af forholdet mellem direkte lyd og efterklangsforholdet i et lydsignal |
CN112509594A (zh) * | 2020-06-22 | 2021-03-16 | 中兴通讯股份有限公司 | 一种终端、发声方法、存储介质及电子装置 |
US11290814B1 (en) | 2020-12-15 | 2022-03-29 | Valeo North America, Inc. | Method, apparatus, and computer-readable storage medium for modulating an audio output of a microphone array |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20040019362A (ko) * | 2001-07-20 | 2004-03-05 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | 후처리기로서 멀티 마이크로폰 에코 억제기를 가지는 음향보강 시스템 |
KR20050115857A (ko) * | 2002-12-11 | 2005-12-08 | 소프트맥스 인코퍼레이티드 | 안정성 강제하에서 독립 성분 분석을 사용하여 음향을처리하는 시스템 및 방법 |
KR20060061259A (ko) * | 2004-12-01 | 2006-06-07 | 하만 벡커 오토모티브 시스템스 - 웨이브마커 인크. | 잔향 추정 및 억제 시스템 |
KR20100010136A (ko) * | 2008-07-22 | 2010-02-01 | 삼성전자주식회사 | 잡음을 제거하는 장치 및 방법 |
KR20120054087A (ko) * | 2009-09-07 | 2012-05-29 | 퀄컴 인코포레이티드 | 멀티채널 신호의 반향 제거를 위한 시스템, 방법, 장치 및 컴퓨터 판독가능 매체 |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005006808A1 (en) * | 2003-07-11 | 2005-01-20 | Cochlear Limited | Method and device for noise reduction |
US8271277B2 (en) | 2006-03-03 | 2012-09-18 | Nippon Telegraph And Telephone Corporation | Dereverberation apparatus, dereverberation method, dereverberation program, and recording medium |
EP2058804B1 (en) | 2007-10-31 | 2016-12-14 | Nuance Communications, Inc. | Method for dereverberation of an acoustic signal and system thereof |
WO2009110578A1 (ja) | 2008-03-03 | 2009-09-11 | 日本電信電話株式会社 | 残響除去装置、残響除去方法、残響除去プログラム、および記録媒体 |
US8265290B2 (en) | 2008-08-28 | 2012-09-11 | Honda Motor Co., Ltd. | Dereverberation system and dereverberation method |
US8989360B2 (en) * | 2011-03-04 | 2015-03-24 | Mitel Networks Corporation | Host mode for an audio conference phone |
-
2014
- 2014-10-07 KR KR1020177008048A patent/KR20170063618A/ko active IP Right Grant
- 2014-10-07 WO PCT/KR2014/009405 patent/WO2016056683A1/ko active Application Filing
- 2014-10-07 US US15/517,855 patent/US9997170B2/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20040019362A (ko) * | 2001-07-20 | 2004-03-05 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | 후처리기로서 멀티 마이크로폰 에코 억제기를 가지는 음향보강 시스템 |
KR20050115857A (ko) * | 2002-12-11 | 2005-12-08 | 소프트맥스 인코퍼레이티드 | 안정성 강제하에서 독립 성분 분석을 사용하여 음향을처리하는 시스템 및 방법 |
KR20060061259A (ko) * | 2004-12-01 | 2006-06-07 | 하만 벡커 오토모티브 시스템스 - 웨이브마커 인크. | 잔향 추정 및 억제 시스템 |
KR20100010136A (ko) * | 2008-07-22 | 2010-02-01 | 삼성전자주식회사 | 잡음을 제거하는 장치 및 방법 |
KR20120054087A (ko) * | 2009-09-07 | 2012-05-29 | 퀄컴 인코포레이티드 | 멀티채널 신호의 반향 제거를 위한 시스템, 방법, 장치 및 컴퓨터 판독가능 매체 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017218128A1 (en) * | 2016-06-15 | 2017-12-21 | Intel Corporation | Far field automatic speech recognition pre-processing |
US10657983B2 (en) | 2016-06-15 | 2020-05-19 | Intel Corporation | Automatic gain control for speech recognition |
Also Published As
Publication number | Publication date |
---|---|
KR20170063618A (ko) | 2017-06-08 |
US20170309294A1 (en) | 2017-10-26 |
US9997170B2 (en) | 2018-06-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2016056683A1 (ko) | 전자 장치 및 이의 잔향 제거 방법 | |
EP3531674B1 (en) | Sound processing method and device | |
US20180070173A1 (en) | Methods circuits devices systems and associated computer executable code for acquiring acoustic signals | |
US10827263B2 (en) | Adaptive beamforming | |
WO2018008885A1 (ko) | 영상처리장치, 영상처리장치의 구동방법 및 컴퓨터 판독가능 기록매체 | |
US9672821B2 (en) | Robust speech recognition in the presence of echo and noise using multiple signals for discrimination | |
US20120249785A1 (en) | Signal processor and signal processing method | |
EP1489596B1 (en) | Device and method for voice activity detection | |
WO2012161555A2 (ko) | 방향성 마이크 어레이를 이용한 신호 분리시스템 및 그 제공방법 | |
CN110770827B (zh) | 基于相关性的近场检测器 | |
US20090147942A1 (en) | Reducing Echo | |
US8363846B1 (en) | Frequency domain signal processor for close talking differential microphone array | |
KR20070004893A (ko) | 적응성 빔 형성기, 사이드로브 소거기, 핸즈프리 음성 통신디바이스 | |
CN103238182A (zh) | 具有远程噪声检测器的降噪系统 | |
CN111435598B (zh) | 语音信号处理方法、装置、计算机可读介质及电子设备 | |
GB2519379A (en) | Noise reduction in multi-microphone systems | |
EP4044181A1 (en) | Deep learning speech extraction and noise reduction method fusing signals of bone vibration sensor and microphone | |
CN110830870B (zh) | 一种基于传声器技术的耳机佩戴者语音活动检测系统 | |
EP3545691B1 (en) | Far field sound capturing | |
CN112363112B (zh) | 一种基于线性麦克风阵列的声源定位方法及装置 | |
US9729967B2 (en) | Feedback canceling system and method | |
KR20110021306A (ko) | 마이크로폰의 신호 보상 장치 및 그 방법 | |
CN113223544A (zh) | 音频的方向定位侦测装置及方法以及音频处理系统 | |
El Baba et al. | Time of arrival disambiguation using the linear Radon transform | |
US20190035382A1 (en) | Adaptive post filtering |
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: 14903521 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 20177008048 Country of ref document: KR Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15517855 Country of ref document: US |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 14903521 Country of ref document: EP Kind code of ref document: A1 |