A method for measuring room acoustic reflection boundaries

Abstract

The present invention relates to a method for measuring the acoustic reflection boundary of a room. The room impulse response of each loudspeaker relative to each microphone is measured; the TDOA value of the first reflected sound signal is calculated; the image sound source coordinates and the deviation and covariance matrix of the coordinates are calculated; the position of the acoustic reflection boundary is calculated according to the image sound source coordinates and the deviation and covariance matrix of the coordinates; wherein, all combinations of the TDOA values ​​of any loudspeaker relative to the first reflected sound signals of any two microphones are extracted according to the TDOA values ​​of all first reflected sound signal combinations; the image sound source coordinates of the loudspeaker and the deviation and covariance matrix of the image sound source coordinates are calculated; the above operation is performed on each loudspeaker to obtain the image sound source coordinates of all loudspeakers and the deviation and covariance matrix of the image sound source coordinates. The present invention has high positioning accuracy and small calculation amount, and can perform real-time positioning.

Description

Technical Field

[0001] The invention relates to a method for measuring room acoustic reflection boundaries. Background Art

[0002] In many fields of audio signal processing, acoustic reflection boundary positioning plays an important role in improving computing power, such as sound source positioning, sound field reconstruction, audio forensics, etc. In addition, the acquisition of room geometry model parameters can also enhance the target signal in the sound field and be applied in computing methods such as speech enhancement, sound source separation, and dereverberation.

[0003] Traditionally, there are two methods for locating the specific position of the acoustic reflection boundary, one is the image sound source method and the other is the ellipse method. The image sound source method first uses TDOA (Time difference of arrival) to locate the image sound source obtained after the actual sound source is reflected by the room, and then determines the reflection boundary based on the position of the actual sound source and the image sound source. The ellipse method constructs a nonlinear optimization problem based on the conditions satisfied by the reflected sound TOA (Time of arrival), and directly locates the reflection boundary by solving the optimization problem. The image sound source method has a fast calculation speed, but low accuracy; the ellipse method has a slow calculation speed, but is more accurate.

[0004] In the measurement of room reflection boundaries, due to the delay between the transmitter of the acoustic signal and the receiver of the signal, the propagation time of the measured signal itself will introduce an error, which is hereinafter referred to as the delay error. The traditional image sound source method uses TDOA information for positioning, which can avoid delay errors. At the same time, this method can also obtain a closed-form solution with a fast calculation speed. However, since this method needs to calculate the coordinates of the image sound source first and then calculate the position of the reflection boundary, the error of the image sound source coordinate positioning will be amplified, and the final result will still have a large deviation. The ellipse line method uses TOA information for positioning, which itself will introduce delay errors, and this method requires nonlinear optimization, has a slow calculation speed, and is difficult to achieve real-time measurement.

[0005] In addition, the traditional method of calculating TDOA or TOA is to first use a swept frequency signal to measure the room impulse response of each sound source one by one. This measurement method will introduce errors due to the asynchrony of each channel, hereinafter referred to as synchronization error. Summary of the invention

[0006] In order to solve the above technical defects, the present invention provides a sound effect processing system and method.

[0007] The technical solution to the above technical problem is: a method for measuring the acoustic reflection boundary of a room, comprising the following steps:

[0008] S1: Measure the room impulse response of each loudspeaker relative to each microphone;

[0009] S2: Calculate the TDOA value of the first reflected sound signal;

[0010] S3: Calculate the image source coordinates and the coordinate deviation and covariance matrix;

[0011] S4: Calculate the position of the acoustic reflection boundary according to the image sound source coordinates and the coordinate deviation and covariance matrix;

[0012] The specific process of step S3 is as follows:

[0013] S3.1 extracting all combinations of TDOA values ​​of the first reflection sound signals of any loudspeaker relative to any two microphones from the TDOA values ​​of all combinations of the first reflection sound signals obtained in step S2;

[0014] S3.2 calculate the image sound source coordinates of the speaker and the deviation and covariance matrix of the image sound source coordinates;

[0015] S3.3 Perform the above operation on each speaker to obtain the image source coordinates of all speakers and the deviation and covariance matrix of the image source coordinates; let the covariance matrix of the image source coordinate numbered m be cov m , the three non-repeating elements of the matrix are n xx,m 、n xy,m 、n yy,m , then the covariance matrix is ​​as follows:

[0016]

[0017] Among them, n xx,m is the covariance between x coordinates and x coordinates, n xy,m is the covariance of the x-coordinate and the y-coordinate, n yy,m The covariance of the y-coordinate and the y-coordinate is .

[0018] Further, the specific process of step S4 is:

[0019] S4.1 calculates the following parameters in sequence based on the image source coordinates and the known loudspeaker coordinates:

[0020]

[0021] E xx =diag{n xx,1 ,n xx,2 ,…,n xx,M}

[0022] E xy =diag{n xy,1 ,nxy,2 ,…,n xy,M}

[0023] E yy =diag{n yy,1 ,n yy,2 ,…,n yy,M}

[0024]

[0025]

[0026]

[0027]

[0028]

[0029]

[0030] Among them, x S,1 、x S,2 ……x S,M and S,1 ,y S,2 ……y S,M It represents the x-axis coordinate and y-axis coordinate of the speakers numbered 1, 2, ..., M in the rectangular coordinate system among all the M speakers. I,1 、x I,2 ……x I,M and I,1 ,y I,2 ……y I,M It represents the x-axis coordinate and y-axis coordinate of the image sound source corresponding to the speakers numbered 1, 2, ..., M in the rectangular coordinate system among all the M speakers. xx,1 ,n xx,2 ,…,n xx,M} means that the main diagonal elements are n xx,1 ,n xx,2 ,…,n xx,Mxx , a diagonal matrix with the remaining elements being 0. The upper right subscript T indicates the transpose of the matrix, and the lower right subscript 2M×2 indicates that the matrix has 2M rows and 2 columns; n xx,1 ,n xx,2 ,…,n xx,M 、n xy,1 ,n xy,2 ,…,n xy,M 、n yy,1 ,n yy,2 ,…,n yy,M is the element in the covariance matrix of each image source coordinate;

[0031] S4.2 Assume that the equation of the acoustic reflection boundary is l1x+l2y+1=0, where l1 and l2 are the linear parameters of the acoustic reflection boundary equation. Calculate the preliminary estimate of the linear parameters of the acoustic reflection boundary equation.

[0032]

[0033] in, is a preliminary estimate of the linear parameters of the acoustic reflection boundary equation, and the parameter b is

[0034]

[0035] S4.3 Based on the preliminary estimate of the linear parameters of the acoustic reflection boundary equation, recalculate the parameter Ψ as

[0036]

[0037]

[0038]

[0039]

[0040] S4.4 Calculate the final estimate of the linear parameters of the acoustic reflection boundary equation

[0041] θ=(A T Ψ -1 A) -1 A T Ψ -1 b

[0042] in, is the final estimate of the linear parameters of the acoustic reflection boundary equation.

[0043] Furthermore, the room acoustic reflection boundary measurement method further includes: step S5: correcting the measurement result according to the final estimated value of the linear parameter of the acoustic reflection boundary equation, the specific process is as follows:

[0044] S5.1 Calculation parameter matrix

[0045] E x =[n x,1 ,n x,2 ,…,n x,M ] T

[0046] E y =[n y,1 ,n y,2 ,…,n y,M ] T

[0047]

[0048] Among them, n x,1 ,n x,2 ,…,n x,M and n y,1 ,n y,2 ,…,n y,M is the deviation of the x-coordinate and the y-coordinate of each image sound source coordinate, obtained by the deviation of the image sound source coordinate calculated in step S3;

[0049] S5.2 Calculate the corrected values ​​of the linear parameters of the acoustic reflection boundary equation

[0050]

[0051] in, is the correction value of the linear parameter of the acoustic reflection boundary equation. The parameters θ, A, and Ψ are calculated in step S4.

[0052] Further, the specific process of step S1 is:

[0053] S1.1 generates a coded signal and a decoded signal;

[0054] S1.2 feeds the generated coded signal to all speakers simultaneously for playback;

[0055] S1.3 use the microphone array to receive the signal played by the speaker array to obtain a measurement signal;

[0056] S1.4 calculates the periodic cross-correlation between the measured signal and the decoded signal obtained by each microphone, and obtains the room impulse response of each loudspeaker relative to each microphone.

[0057] Further, the specific process of step S2 is:

[0058] S2.1 extracting the first reflection sound signal of the room impulse response according to the measured room impulse response;

[0059] S2.2 calculates the generalized cross-correlation function between any two first reflection sound signals;

[0060] S2.3 obtaining a time delay value corresponding to the maximum value of the generalized cross-correlation function, where the time delay value is the calculated TDOA value of the two first reflected sound signals;

[0061] S2.4 Repeat the above S2.1-S2.3 operations for all pairs of non-repeating first reflection sound signal combinations to obtain the TDOA values ​​of all first reflection sound signal combinations.

[0062] Compared with the prior art, the present invention has the following significant advantages: the present invention obtains higher positioning accuracy by improving the acoustic reflection boundary positioning algorithm and correcting the positioning result by estimating the deviation of the positioning result. Compared with the traditional ellipse positioning method, the present invention avoids the influence of delay error by using TDOA for calculation; and obtains a faster calculation speed by providing a closed-form solution of the final positioning result. In addition, compared with the two traditional methods, the present invention avoids the influence of synchronization error by simultaneously measuring the room impulse response of each speaker relative to each microphone. In summary, the present invention has the advantages of high positioning accuracy, small calculation amount, and can perform real-time positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a simplified flow chart of the steps of the present invention;

[0064] Figure 2 It is a detailed diagram of the sound process of the present invention. DETAILED DESCRIPTION

[0065] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, a person skilled in the art can imagine various implementation methods of the present invention. Therefore, the following specific implementation methods and drawings are only exemplary descriptions of the technical solution of the present invention, and should not be regarded as the whole of the present invention or as a limitation or restriction of the technical solution of the present invention.

[0066] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that the relative arrangement of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless otherwise specifically stated.

[0067] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.

[0068] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered as part of the specification.

[0069] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0070] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0071] like Figure 1As shown, the specific implementation process of the room acoustic reflection boundary measurement method of the present invention is as follows:

[0072] Step S1: Measure the room impulse response

[0073] To avoid introducing synchronization errors, the present invention simultaneously measures the room impulse response of each speaker relative to each microphone. In this step,

[0074] S1.1, first generate a coded signal and a decoded signal;

[0075] S1.2 feeds the generated encoded signal to all speakers for playback simultaneously;

[0076] S1.3 using a microphone array to receive a measurement signal, that is, using a microphone array to receive a signal played by a loudspeaker array;

[0077] S1.4, finally, the periodic cross-correlation between the measured signal and the decoded signal obtained by each microphone is calculated to obtain the room impulse response of each loudspeaker relative to each microphone.

[0078] The method for generating the coded signal and the decoded signal can adopt the existing technology, such as the method described in the document "Multi-source room impulse response measurement method based on ternary pseudo-random sequence set" [J / OL]. Applied Acoustics: 1-18.

[0079] Step S2: Calculate the TDOA of the first reflected sound signal

[0080] The present invention uses the generalized cross-correlation method to calculate the TDOA of the first reflected sound. The calculation process is as follows:

[0081] S2.1 First, based on the measured room impulse response, extract the first reflection sound part of the room impulse response, hereinafter referred to as the first reflection sound signal, and the number of the first reflection sound signals is the same as the number of the room impulse response;

[0082] S2.2 calculates the generalized cross-correlation function between any two first reflection sound signals;

[0083] S2.3 finding the time delay value corresponding to the maximum value of the generalized cross-correlation function, where the time delay value is the TDOA estimation value of the two first reflected sound signals calculated;

[0084] S2.4 Finally, the above S2.1-S2.3 operations are repeated for all pairs of non-repeating first reflection sound signal combinations to obtain the TDOA estimation values ​​of all first reflection sound signal combinations.

[0085] In this step, the first reflection part of the room impulse response can be extracted using existing technology, such as the method described in the literature Remaggi L, Jackson PJB, Coleman P, et al. Acoustic reflector localization: novel image source reversion and direct localization methods [J]. IEEE / ACM Transactions on Audio, Speech, and Language Processing, 2016, 25 (2): 296-309. The generalized cross-correlation function between any two first reflection sound signals can also be calculated using existing technology, such as the literature Tervo S, The method described in J,Lokki T.Acoustic reflection localization from room impulse responses[J].ACTA Acustica united with Acustica,2012,98(3):418-440.

[0086] Step S3: Calculate the image source coordinates and the deviation and covariance matrix of the coordinates

[0087] The image sound source refers to the actual sound-generating loudspeaker, which is a mirror image symmetrical to the acoustic reflection boundary axis. The specific process of calculating the image sound source coordinates and the coordinate deviation and covariance matrix of the present invention is as follows:

[0088] S3.1 extract all combinations of TDOA estimates of first reflection sound signals of a certain loudspeaker relative to any two microphones from the TDOA estimates of all combinations of first reflection sound signals obtained in the previous step;

[0089] S3.2 calculate the image sound source coordinates of the speaker and the deviation and covariance matrix of the image sound source coordinates;

[0090] S3.3 Perform the above operation on each speaker to obtain the image source coordinates of all speakers and the deviation and covariance matrix of the image source coordinates. The covariance matrix of each image source coordinate calculated should be a 2×2 matrix with two identical elements. Let the covariance matrix of the image source coordinate numbered m be cov m , the three non-repeating elements of the matrix are n xx,m 、n xy,m 、n yy,m , then the covariance matrix is ​​as follows:

[0091]

[0092] Among them, n xx,m is the covariance between x coordinates and x coordinates, n xy,m is the covariance of the x-coordinate and the y-coordinate, n yy,m is the covariance between the y coordinates and the y coordinates.

[0093] Assume n x,m is the deviation of the x-coordinate of the image source numbered m, n y,m is the deviation of the y coordinate of the image sound source numbered m.

[0094] The method for calculating the image sound source coordinates of the speaker and the deviation and covariance matrix of the image sound source coordinates can be the method in the document Ho K C.Bias reduction for an explicit solution of source localizationusing TDOA[J].IEEE Transactions on Signal Processing, 2012, 60(5):2101-2114.

[0095] Step S4: Locate acoustic reflection boundaries

[0096] Using the image sound source coordinates calculated in the previous step as well as the coordinate deviation and covariance matrix, calculate the position of the acoustic reflection boundary. The specific calculation process is as follows:

[0097] S4.1 Calculate the following parameters in sequence based on the image source coordinates and the known speaker coordinates

[0098]

[0099] E xx =diag{n xx,1 ,n xx,2 ,…,n xx,M}

[0100] E xy =diag{n xy,1 ,n xy,2 ,…,n xy,M}

[0101] E yy =diag{n yy,1 ,n yy,2 ,…,n yy,M}

[0102]

[0103]

[0104]

[0105]

[0106]

[0107]

[0108] Among them, x S,1 、x S,2 ……x S,M and S,1 ,y S,2 ……y S,M It represents the x-axis coordinate and y-axis coordinate of the speakers numbered 1, 2, ..., M in the rectangular coordinate system among all the M speakers. I,1 、x I,2 ……x I,M and I,1 ,y I,2 ……y I,M It represents the x-axis coordinate and y-axis coordinate of the image sound source in the rectangular coordinate system corresponding to the speakers numbered 1, 2, ..., M among all the M speakers. diag{a,b,c...} represents a diagonal matrix whose main diagonal elements are a,b,c... and the remaining elements are 0. The upper right subscript T represents the transpose of the matrix, and the lower right subscript a×b represents that the matrix is ​​an a×b matrix. That is, diag{n xx,1 ,n xx,2 ,…,n xx,M} means the main diagonal elements are A diagonal matrix with all other elements being zero. xx,1 ,n xx,2 ,…,n xx,M 、n xy,1 ,n xy,2 ,…,n xy,M 、n yy,1 ,n yy,2 ,…,n yy,M is the element in the covariance matrix of each image sound source coordinate, which is obtained from the covariance matrix of the image sound source coordinate calculated in the previous step.

[0109] S4.2 Assume that the equation of the acoustic reflection boundary is l1x+l2y+1=0, where l1 and l2 are the linear parameters of the acoustic reflection boundary equation. Calculate the preliminary estimate of the linear parameters of the acoustic reflection boundary equation.

[0110]

[0111] in, is a preliminary estimate of the linear parameters of the acoustic reflection boundary equation, and the parameter b is

[0112]

[0113] S4.3 Based on the preliminary estimate of the linear parameters of the acoustic reflection boundary equation, recalculate the parameter Ψ as

[0114]

[0115]

[0116]

[0117]

[0118] S 4.4 Calculate the final estimate of the linear parameters of the acoustic reflection boundary equation

[0119] θ=(A T Ψ -1 A) -1 A T Ψ -1 b

[0120] in, It is the final estimate of the linear parameters of the acoustic reflection boundary equation. That is, the equation of the acoustic reflection boundary is l1x+l2y+1=0. This equation determines the position of the acoustic reflection boundary.

[0121] Step S5: Correct the positioning result

[0122] The measurement results are corrected according to the final estimated values ​​of the linear parameters of the acoustic reflection boundary equation. The specific process is as follows:

[0123] S5.1 Calculation parameter matrix

[0124] E x =[n x,1 ,n x,2 ,…,n x,M ] T

[0125] E y =[n y,1 ,n y,2 ,…,n y,M ] T

[0126]

[0127] Among them, n x,1 ,n x,2 ,…,n x,M and n y,1 ,n y,2 ,…,ny,M is the deviation of the x-coordinate and y-coordinate of each image source coordinate, obtained by the deviation of the image source coordinate calculated in step 3.

[0128] S5.2 Calculate the corrected values ​​of the linear parameters of the acoustic reflection boundary equation

[0129]

[0130] in, is the correction value of the linear parameter of the acoustic reflection boundary equation. The parameters θ, A, and Ψ are calculated in step S4.

Claims

1. A method for measuring room acoustic reflection boundaries, characterized in that: The steps include: S1: Measure the room impulse response of each loudspeaker relative to each microphone; S2: Calculate the TDOA value of the first reflected sound signal; S3: Calculate the image sound source coordinates and the deviation and covariance matrix of the image sound source coordinates; S4: Calculate the position of the acoustic reflection boundary according to the image sound source coordinates and the deviation and covariance matrix of the image sound source coordinates; The specific process of step S3 is as follows: S3.1 extracting all combinations of TDOA values ​​of the first reflection sound signals of any loudspeaker relative to any two microphones from the TDOA values ​​of all combinations of the first reflection sound signals obtained in step S2; S3.2 calculate the image sound source coordinates of the speaker and the deviation and covariance matrix of the image sound source coordinates; S3.3 Perform the above operation on each speaker to obtain the image source coordinates of all speakers and the deviation and covariance matrix of the image source coordinates; let the covariance matrix of the image source coordinate numbered m be cov m The three non-repeating elements of the covariance matrix are n xx,m 、n xy,m 、n yy,m , then the covariance matrix is ​​as follows: Among them, n xx,m is the covariance between the x-coordinate and the x-coordinate, n xy,m is the covariance of the x-coordinate and the y-coordinate, n yy,m is the covariance between the y coordinates and the y coordinates; The specific process of step S4 is: S4.1 calculates the following parameters in sequence based on the image source coordinates and the known loudspeaker coordinates: Yes xx =diag{n xx,1 ,n xx,2 ,…,n xx,M } Yes xy =diag{n xy,1 ,n xy,2 ,…,n xy,M } Yes yy =diag{n yy,1 ,n yy,2 ,…,n yy,M } Among them, x S,1 、x S,2 , …, x S,M and S,1 ,y S,2 , …, y S,M represents the x-axis coordinates and y-axis coordinates of the speakers numbered 1, 2, ..., M in the rectangular coordinate system among all the M speakers; I,1 、x I,2 , …, x I,M and I,1 ,y I,2 , …, y I,M represents the x-axis coordinate and y-axis coordinate of the image sound source corresponding to the speakers numbered 1, 2, ..., M in the rectangular coordinate system among all M speakers; diag{n xx,1 ,n xx,2 ,…,n xx,M } means that the main diagonal elements are n xx,1 ,n xx,2 ,…,n xx,Mxx , a diagonal matrix with the remaining elements being 0; T represents the transpose of the matrix, 2M×2 represents a matrix with 2M rows and 2 columns; n xx,1 ,n xx,2 ,…,n xx,M 、n xy,1 ,n xy,2 ,…,n xy,M 、n yy,1 ,n yy,2 ,…,n yy,M is the element in the covariance matrix of each image source coordinate; S4.2 Assume that the equation of the acoustic reflection boundary is l1x+l2y+1=0, where l1 and l2 are the linear parameters of the acoustic reflection boundary equation. Calculate the preliminary estimate of the linear parameters of the acoustic reflection boundary equation. in, is a preliminary estimate of the linear parameters of the acoustic reflection boundary equation, and the parameter b is S4.3 Based on the preliminary estimate of the linear parameters of the acoustic reflection boundary equation, recalculate the parameter Ψ as S 4.4 Calculate the final estimate of the linear parameters of the acoustic reflection boundary equation θ=(A T P -1 A) -1 A T P -1 b in, is the final estimate of the linear parameters of the acoustic reflection boundary equation.

2. The room acoustic reflection boundary measurement method according to claim 1, characterized in that: Also includes: Step S5: Correct the measurement result according to the final estimated value of the linear parameter of the acoustic reflection boundary equation. The specific process is as follows: S5.1 Calculation parameter matrix IT x [n x,1 ,n x,2 ,…,n x,M ] T IT y [n y,1 ,n y,2 ,…,n y,M ] T Among them, n x,1 ,n x,2 ,…,n x,M and n y,1 ,n y,2 ,…,n y,M is the deviation of the x-coordinate and the y-coordinate of each image sound source coordinate, obtained by the deviation of the image sound source coordinate calculated in step S3; S5.2 Calculate the corrected values ​​of the linear parameters of the acoustic reflection boundary equation in, is the correction value of the linear parameter of the acoustic reflection boundary equation. The parameters θ, A, and Ψ are calculated in step S4.

3. The room acoustic reflection boundary measurement method according to claim 1, characterized in that: The specific process of step S1 is: S1.1 generates a coded signal and a decoded signal; S1.2 feeds the generated coded signal to all speakers simultaneously for playback; S1.3 use the microphone array to receive the signal played by the speaker array to obtain a measurement signal; S1.4 calculates the periodic cross-correlation between the measured signal and the decoded signal obtained by each microphone, and obtains the room impulse response of each loudspeaker relative to each microphone.

4. The room acoustic reflection boundary measurement method according to claim 1, characterized in that: The specific process of step S2 is: S2.1 extracting the first reflection sound signal of the room impulse response according to the measured room impulse response; S2.2 calculates the generalized cross-correlation function between any two first reflection sound signals; S2.3 obtaining a time delay value corresponding to the maximum value of the generalized cross-correlation function, where the time delay value is the calculated TDOA value of the two first reflected sound signals; S2.4 Repeat the above S2.1-S2.3 operations for all pairs of non-repeating first reflection sound signal combinations to obtain the TDOA values ​​of all first reflection sound signal combinations.

Images