WO2015129977A1 - Procédé et dispositif pour détecter un emplacement de source sonore par utilisation d'une moyenne pondérée selon la précision de chaque bande - Google Patents

Procédé et dispositif pour détecter un emplacement de source sonore par utilisation d'une moyenne pondérée selon la précision de chaque bande Download PDF

Info

Publication number
WO2015129977A1
WO2015129977A1 PCT/KR2014/008720 KR2014008720W WO2015129977A1 WO 2015129977 A1 WO2015129977 A1 WO 2015129977A1 KR 2014008720 W KR2014008720 W KR 2014008720W WO 2015129977 A1 WO2015129977 A1 WO 2015129977A1
Authority
WO
WIPO (PCT)
Prior art keywords
sound source
weighted average
band
source position
snr
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/KR2014/008720
Other languages
English (en)
Korean (ko)
Inventor
송재종
박성주
양창모
박호종
구자성
한택진
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Korea Electronics Technology Institute
Original Assignee
Korea Electronics Technology Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Korea Electronics Technology Institute filed Critical Korea Electronics Technology Institute
Publication of WO2015129977A1 publication Critical patent/WO2015129977A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S3/00—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic or electromagnetic waves, or particle emission, not having a directional significance, are being received
    • G01S3/80—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic or electromagnetic waves, or particle emission, not having a directional significance, are being received using ultrasonic, sonic or infrasonic waves

Definitions

  • the present invention relates to a method and apparatus for detecting a sound source position having a direction through a panning technique in a music signal in which several musical instrument sources including vocals are mixed.
  • position estimation in a stereo sound source estimates a sound source position of a primary component having a direction.
  • the sound source position of the main component is shaken.
  • the main component is separated using the PCA and the sound source position of the main component is estimated.
  • the sound source position is estimated for each band, the estimated sound source position is converted into one frame position to detect the actual sound source position.
  • the accuracy of the sound source position estimated for each band is different, there is a problem that the accuracy is reduced when estimating the frame position using a uniform average.
  • the surrounding components have more energy at low frequencies, reducing the accuracy of the low band estimated sound source position. This is because both the main component and the surrounding components decrease in energy at higher frequencies.
  • the sound source position estimated in the low energy section is not accurate, and thus, when the frame position is estimated using a uniform average without considering accuracy, the accuracy of the detected sound source position is lowered.
  • the prior art does not efficiently estimate the accuracy of the position of the correct sound source object in the music signal including several sound sources.
  • An object of the present invention is to provide an apparatus and method for improving the accuracy of sound source object position estimation efficiently and accurately.
  • another object of the present invention is to determine the high accuracy interval of the estimated sound source position in the estimation of the sound source position, the high accuracy interval can be obtained using the estimated SNR and energy for each band using the PCA, accuracy It is aimed at estimating the position of a sound source more precisely through a weighted average.
  • the input stereo signal is separated into a main component and a peripheral component by a PCA (Principle Component Analysis) technique and an eigenvector for the main component of the stereo signal. Extracting; Extracting a sound source position for each band of the stereo signal using the extracted eigenvectors; And calculating a weighted average according to the accuracy of the extracted sound source position using energy of each band of the stereo signal.
  • PCA Principal Component Analysis
  • the method for detecting a sound source position using a weighted average according to the accuracy of each band may further include detecting a frame sound source position of the stereo signal using the calculated weighted average.
  • the calculating of the weighted average may include determining whether to calculate a weighted average according to the energy of each band; And calculating a weighted average using SNR for the band in which the calculation of the weighted average is determined.
  • the calculating of the weighted average using the SNR may include calculating an estimated SNR using an estimated original signal and a peripheral component signal separated by the PCA technique, and using the calculated estimated SNR to weight the frame sound source position. It is preferable to calculate the average.
  • the sound source position detection apparatus using a weighted average according to the present embodiment for solving the technical problem is to separate the input stereo signal into a principal component and a peripheral component by the PCA (Principle Component Analysis) technique to the eigenvector for the principal component of the stereo signal
  • a vector extraction unit for extracting the
  • a band-specific sound source position extracting unit for extracting a band-specific sound source position of the stereo signal using the extracted eigenvectors
  • a weighted average calculator configured to calculate a weighted average according to the accuracy of the extracted sound source position using energy of each band of the stereo signal
  • a frame sound source position detector for detecting a frame sound source position of the stereo signal using the calculated weighted average.
  • the position estimation accuracy of the sound source object can be improved, the position of the sound source object in the next source music signal can be more accurately implemented. It is possible to effectively improve the sound source object position estimation accuracy by using.
  • the sound source object may be rearranged.
  • FIG. 1 is a block diagram illustrating a sound source position detection apparatus using a weighted average according to an embodiment of the present invention.
  • FIG. 2 is a flowchart illustrating a sound source position detection method using a weighted average according to an embodiment of the present invention.
  • FIG 3 is an exemplary view showing an estimated sound source position error according to SNR of the sound source position detection method using a weighted average according to an embodiment of the present invention.
  • FIG. 4 is an exemplary diagram illustrating weights according to SNR of a sound source position detection method using a weighted average according to an embodiment of the present invention.
  • FIG. 5 is an exemplary diagram illustrating an estimated sound source position error according to energy of a sound source position detection method using a weighted average according to an embodiment of the present invention.
  • FIG. 1 is a block diagram illustrating a sound source position detection apparatus using a weighted average according to band-specific accuracy according to an embodiment of the present invention.
  • the sound source position detection apparatus using the weighted average according to the accuracy of each band according to the present embodiment includes a vector extractor 100, a sound source extractor 200 for each band, a weighted average calculator 300, and a sound source position detector.
  • the vector extractor 100 extracts an eigenvector for the main component of the stereo signal by separating the input stereo signal into a main component and a peripheral component by a PCA (Principle Component Analysis) technique.
  • PCA Principal Component Analysis
  • the band-specific sound source extractor 200 extracts the band-specific sound source positions of the stereo signal using the extracted eigenvectors.
  • the weighted average calculator 300 calculates a weighted average according to the accuracy of the extracted sound source position using energy of each band of the stereo signal.
  • the sound source position detector detects the frame sound source position of the stereo signal using the calculated weighted average.
  • the weighted average calculation unit 300 calculates a weighted average using the SNR for the energy weighted average calculation unit 300 for determining whether to calculate the weighted average according to the energy for each band, and the band for which the calculation of the weighted average is determined
  • the SNR weighted average calculation unit 300 is further included.
  • the SNR weighted average calculation unit 300 calculates an estimated SNR by using the estimated original signal and the peripheral component signals separated by the PCA method, and the calculated estimation. The SNR is used to calculate a weighted average for detection of the frame sound source position.
  • the sound source position detector detects the frame sound source position of the stereo signal using the calculated weighted average.
  • FIG. 2 is a flowchart illustrating a sound source position detection method using a weighted average according to band-specific accuracy.
  • the sound source position detection method using a weighted average according to the accuracy of each band according to the present embodiment is a vector extraction step (S100), a band-specific sound source extraction step (S200), a weighted average calculation step (S300), a sound source Position detection step (S400).
  • the vector extractor extracts the eigenvectors for the main components of the stereo signal by separating the main components and the peripheral components by the PCA (Principle Component Analysis) technique.
  • PCA Principal Component Analysis
  • eigenvectors first, for extracting the eigenvectors, first perform N-point Discrete Fourier Transform (DFT) per frame of the input stereo signal, and then obtain covariance in the frequency domain.
  • DFT Discrete Fourier Transform
  • the eigenvectors of the left and right signals are estimated after separating the main and peripheral components using the PCA technique.
  • the band-specific sound source extractor 200 estimates the panning gain for each band using the estimated eigenvectors and converts the panning gain to the sound source position to extract the sound source position for each band of the stereo signal. do.
  • the weighted average calculator 300 calculates a weighted average according to the accuracy of the sound source position extracted by using band-specific energy of the stereo signal.
  • a method of calculating the weighted average according to the present embodiment will be described in detail.
  • the weighted average calculation step S300 may extract energy for each band. That is, it is determined whether the weighted average is calculated according to the energy for each band, and the weighted average using the SNR is calculated for the band in which the calculation of the weighted average is determined.
  • the error becomes large and the accuracy decreases for the section where the energy of each band is too low.
  • the sound source position extracted for the section below the random energy is excluded when detecting the frame position through the weighted average or sets a very small weight. To minimize the impact.
  • the weighted average calculation step S300 calculates an estimated SNR weighted average using PCA.
  • the estimated SNR is estimated using the peripheral component signal and the input signal separated by PCA.
  • the weighted average is calculated using the SNR-weight curve obtained through learning.
  • SNR operates in the frequency domain to estimate the SNR using the input signal and the peripheral signal separated by PCA and to obtain the band-specific SNR.
  • the SNR may be extracted in dB scale.
  • P original means power of an original signal
  • P noise means power of a noise signal.
  • SNR is expressed as the ratio of power to the original signal and noise. It is assumed that the signal consisting of the main component is the original signal, and the peripheral component is noise. However, since the information on the original signal is unknown in the sound source position estimation technology, the estimated SNR is used. In this case, the original signal is estimated by separating the main component and the peripheral component of the input signal by using the PCA.
  • X input is the input signal and X primary and X ambient are the main and peripheral components.
  • X ⁇ original is the original signal estimated using the input signal and ambient components
  • X ⁇ ambient is the ambient component separated using PCA. Since the input signal X input has main and peripheral components, the estimated original signal can be obtained by subtracting the X ⁇ ambient extracted from the input signal from the input signal. The estimated SNR is obtained using the surrounding components separated from the estimated original signal and the input signal.
  • FIG. 3 is an exemplary diagram illustrating an estimated sound source position error according to SNR. As the error approaches 0, the sound source position with high accuracy is estimated. In FIG. 3, when the SNR decreases, the error increases, and when the SNR increases, the error decreases. Therefore, the weight corresponding to the SNR value can be estimated using the graph of FIG. 4.
  • FIG. 4 is an exemplary diagram illustrating weights according to SNRs, and is a result calculated by using FIG. 3 as training data. That is, the weight according to the estimated SNR can be obtained using the graph shown in FIG. 4 and the equation for obtaining the weight is as follows.
  • weight is the weight and SNR estimate is the estimated SNR.
  • the reduction in accuracy at higher bands is due to the reduction in energy per band. This is because directionality is difficult to find in the low energy section. Therefore, it is not possible to estimate the sound source position in the low energy section.
  • the energy weighted average may be calculated in parallel with the SNR weighted average. At this time, it is alternatively performed differently from the SNR weighted average. That is, if the energy of a specific band falls below a certain value, the position estimate of the band is not used to extract the frame average.
  • FIG. 5 is an exemplary diagram illustrating an estimated sound source position error according to energy, and an error value is an angle and represents the smallest error at zero.
  • (a) is a graph of the whole energy band, and (b) is an enlarged portion of the energy 0 to 0.1 in the (a) graph.
  • the weighted average is estimated except for a band having an energy of 0.02 or less at the weighted average.
  • the weighted average calculation step (S300) may be performed by the weighted average calculator 300 to calculate a weighted average according to the accuracy of the extracted sound source location using band-specific energy of a stereo signal. After extracting the energy and determining the weighted average, it is possible to perform the SNR weighted average calculation for the low band according to an arbitrary criterion and to perform the alternative energy weighted average calculation for the high band.
  • the sound source position detecting step detects the frame sound source position of the stereo signal using the weighted average calculated by the sound source position detector.
  • the position estimation accuracy of the sound source object can be improved, the position of the sound source object in the next source music signal can be more accurately implemented. It is possible to effectively improve the sound source object position estimation accuracy by using.
  • the sound source object may be rearranged.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Stereophonic System (AREA)
  • Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
  • Auxiliary Devices For Music (AREA)

Abstract

La présente invention porte sur un procédé et un dispositif pour détecter l'emplacement d'une source sonore directionnelle par une technique panoramique à partir d'un signal de musique dans lequel les sources sonores de divers instruments de musique, y compris de la voix, sont mélangées. Le procédé de détection de l'emplacement de la source sonore par utilisation d'une moyenne pondérée, selon la présente invention, comprend les étapes consistant : à séparer un signal stéréo entré en une composante primaire et des composantes ambiantes au moyen d'une analyse en composantes principales (PCA) de manière à extraire un vecteur propre pour la composante primaire du signal stéréo ; à extraire l'emplacement de la source sonore pour chaque bande du signal stéréo à l'aide du vecteur propre extrait ; et à calculer la moyenne pondérée en fonction de la précision de l'emplacement extrait de la source sonore à l'aide de l'énergie de chaque bande de signal stéréo. Selon la présente invention, étant donné que la précision de localisation d'un objet de source sonore peut être améliorée, chaque objet de source sonore est plus précisément localisé dans un signal de musique à multi-source sonore, et la précision de localisation de l'objet de source sonore peut être efficacement améliorée par l'utilisation d'un Rapport Signal/Bruit (SNR) estimé et de l'énergie de chaque bande au lieu d'un procédé de moyenne égalisée utilisé pour des techniques classiques. De plus, les emplacements des objets de source sonore sont estimés pour divers types de musique de façon à permettre le réagencement des objets de source sonore respectifs.
PCT/KR2014/008720 2014-02-27 2014-09-18 Procédé et dispositif pour détecter un emplacement de source sonore par utilisation d'une moyenne pondérée selon la précision de chaque bande Ceased WO2015129977A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR20140023339A KR101509649B1 (ko) 2014-02-27 2014-02-27 대역별 정확도에 따른 가중 평균을 이용한 음원 위치 검출 방법 및 장치
KR10-2014-0023339 2014-02-27

Publications (1)

Publication Number Publication Date
WO2015129977A1 true WO2015129977A1 (fr) 2015-09-03

Family

ID=53032633

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2014/008720 Ceased WO2015129977A1 (fr) 2014-02-27 2014-09-18 Procédé et dispositif pour détecter un emplacement de source sonore par utilisation d'une moyenne pondérée selon la précision de chaque bande

Country Status (2)

Country Link
KR (1) KR101509649B1 (fr)
WO (1) WO2015129977A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108375763A (zh) * 2018-01-03 2018-08-07 北京大学 一种应用于多声源环境的分频定位方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101825949B1 (ko) * 2015-10-06 2018-02-09 전자부품연구원 음원 분리를 포함하는 음원 위치 추정 장치 및 방법
CN119667343A (zh) * 2024-12-23 2025-03-21 广东电网有限责任公司东莞供电局 一种应用于变压器直流偏磁检测的宽带声学成像方法和装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20030046727A (ko) * 2001-12-06 2003-06-18 박규식 서브밴드 씨피에스피 알고리듬을 이용한 음원위치추정방법및 그 시스템
KR100663729B1 (ko) * 2004-07-09 2007-01-02 한국전자통신연구원 가상 음원 위치 정보를 이용한 멀티채널 오디오 신호부호화 및 복호화 방법 및 장치
KR20090050380A (ko) * 2007-11-15 2009-05-20 삼성전자주식회사 오디오 매트릭스 디코딩 방법 및 장치
JP2013507048A (ja) * 2009-10-05 2013-02-28 ハーマン インターナショナル インダストリーズ インコーポレイテッド オーディオ信号の空間的抽出のためのシステム

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20030046727A (ko) * 2001-12-06 2003-06-18 박규식 서브밴드 씨피에스피 알고리듬을 이용한 음원위치추정방법및 그 시스템
KR100663729B1 (ko) * 2004-07-09 2007-01-02 한국전자통신연구원 가상 음원 위치 정보를 이용한 멀티채널 오디오 신호부호화 및 복호화 방법 및 장치
KR20090050380A (ko) * 2007-11-15 2009-05-20 삼성전자주식회사 오디오 매트릭스 디코딩 방법 및 장치
JP2013507048A (ja) * 2009-10-05 2013-02-28 ハーマン インターナショナル インダストリーズ インコーポレイテッド オーディオ信号の空間的抽出のためのシステム

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108375763A (zh) * 2018-01-03 2018-08-07 北京大学 一种应用于多声源环境的分频定位方法
CN108375763B (zh) * 2018-01-03 2021-08-20 北京大学 一种应用于多声源环境的分频定位方法

Also Published As

Publication number Publication date
KR101509649B1 (ko) 2015-04-07

Similar Documents

Publication Publication Date Title
US10602267B2 (en) Sound signal processing apparatus and method for enhancing a sound signal
WO2018111038A1 (fr) Procédé et dispositif d'estimation de temps de réverbération basés sur un microphone multicanal et utilisant un réseau neuronal profond
JP5701164B2 (ja) 位置検出装置及び位置検出方法
EP1701587A2 (fr) Traitement de signaux acoustiques
ATE418124T1 (de) Ausgeglichener objektverfolger in einer bildsequenz
KR20120080409A (ko) 잡음 구간 판별에 의한 잡음 추정 장치 및 방법
WO2018139882A1 (fr) Procédé et appareil de mesure de distance au moyen d'un radar
WO2016056683A1 (fr) Dispositif électronique et son procédé d'élimination de réverbération
KR100877914B1 (ko) 음원위치-지연시간차 상관관계 역 추정에 의한 음원 방향검지 시스템 및 방법
WO2013048029A1 (fr) Procédé de mesure de la position d'un terminal utilisateur
US20160033615A1 (en) Acoustic processing device and acoustic processing method
Li et al. Estimation of relative transfer function in the presence of stationary noise based on segmental power spectral density matrix subtraction
WO2019212069A1 (fr) Procédé de prédiction de structure d'espace intérieur utilisant une analyse de canal de propagation de signal sans fil par apprentissage profond
WO2015129977A1 (fr) Procédé et dispositif pour détecter un emplacement de source sonore par utilisation d'une moyenne pondérée selon la précision de chaque bande
WO2018139887A1 (fr) Procédé et dispositif de configuration adaptative de seuil pour la détection d'objet au moyen d'un radar
WO2011136407A1 (fr) Appareil et procédé de reconnaissance d'image à l'aide d'un appareil photographique stéréoscopique
US20130156221A1 (en) Signal processing apparatus and signal processing method
WO2017030233A1 (fr) Procédé de détection de position par un dispositif informatique mobile, et dispositif informatique mobile l'exécutant
WO2016108406A1 (fr) Appareil et procédé de détection de position de véhicule
WO2012138175A2 (fr) Dispositif sonar destiné à être fixé à une coque et navire équipé de celui-ci
WO2017111386A1 (fr) Appareil d'extraction des paramètres caractéristiques du signal d'entrée, et appareil de reconnaissance de locuteur utilisant ledit appareil
WO2011136405A1 (fr) Dispositif et procédé de reconnaissance d'image à l'aide d'un appareil photographique en trois dimensions (3d)
WO2020171614A1 (fr) Procédé de génération de signal haptique, et dispositif utilisant un modèle de signal audio
WO2023167507A1 (fr) Procédé de traitement de signal et système radar de détection d'être humain
WO2015147363A1 (fr) Procédé destiné à la détermination de la consommation d'alcool par comparaison de la trame de fréquence de signal de différence et support d'enregistrement et dispositif destinés à sa mise en œuvre

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: 14884177

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14884177

Country of ref document: EP

Kind code of ref document: A1