A method for testing performance of a microphone array

Abstract

The application discloses a kind of methods for testing microphone array performance, the method starts host computer and the microphone array to be tested, determines total test number N;Host computer sends start signal to microphone array, host computer passes through pulse control horn's position and speed by initialization parameter, horn and noise source continuously play audio;The probability that the horn's siren model outputs two types is respectively q1 and q2;If the probability of siren sound is large, enter siren positioning procedure, and the position of horn is estimated;Host computer calculates the space rectangular coordinate system coordinates of the position of horn by the position sensor of stepping motor, and the microphone array sends the calculation result to host computer;Error processing is carried out.The test method of the application obtains the error of microphone array by comparing the actual horn position and the calculated horn position, obtains the working cycle by communication interruption, changes the test condition to simulate different environments, and finally obtains the relevant information of microphone array.

Description

Technical Field

[0001] This invention relates to the technical field of smart device testing, and in particular to an apparatus and testing method for testing the performance of a microphone array. Background Technology

[0002] For existing equipment, operating parameters are mainly determined through computer simulation or static testing. However, actual road testing is time-consuming, inconvenient to install, and greatly affected by external factors. Computer simulation has problems such as difficulty in simulating actual road conditions and vehicle movement, resulting in large errors in equipment parameter estimation. Furthermore, static testing has the problem of difficulty in reproducing the impact of vehicle movement on the time-frequency characteristics of signals.

[0003] For microphone array-based horn-hitting detection devices, the main performance metrics include horn recognition rate and horn location accuracy. Current algorithms for horn recognition rate include support vector machine (SVM) models, convolutional neural network (CNN) models, probability graph (PBR) models, and recurrent neural networks (RNNs). Horn location accuracy primarily relies on time delay estimation algorithms, beamforming, and sparse decomposition algorithms. Furthermore, signal-to-noise ratio (SNR) and microphone array structure optimization significantly impact the results. Existing equipment requires on-road installation for practical testing, which is susceptible to challenges due to weather, traffic flow, and other uncontrollable factors.

[0004] Chinese invention patent application CN109254265A discloses a method for locating vehicles honking based on a microphone array, including the following steps: 1) setting up two microphone arrays and obtaining the sound source direction of the honking sound relative to the array in each array; 2) further obtaining the position of the honking vehicle based on the sound source direction obtained in step 1); 3) estimating the time difference of arrival (TDOA) using N microphones; 4) locating the sound source direction of the honking vehicle using a geometric model. The purpose of this technology is to improve the positioning accuracy, reduce the false alarm rate of honking, enhance the robustness of the TDOA estimation, and reduce the computational complexity of the positioning method. However, this technology lacks dynamic simulation and is based on the discrete Fourier transform form, which has high computational complexity. It obtains the TDOA between two devices by solving the peak value of the cross-correlation function, which is not only computationally complex but also requires improvement in accuracy. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a microphone array device testing system that can simulate driving horn blasts in static, dynamic, non-line-of-sight and noisy environments.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] A method for testing the performance of a microphone array includes the following steps:

[0008] Step S1: Start the host computer and the microphone array under test, and determine the total number of tests N; N is a natural number greater than 1; the host computer is installed next to the mobile platform and connected to the microphone array under test on the mobile platform via cables; the speaker of the mobile platform is installed on the connector, and the two ends of the connector are movably connected to two parallel guide rails via sliders; the connector is connected to the stepper motor via a synchronous belt; a limit switch is installed at the end of the guide rail; the guide rail is set on the upper base plate, and a lower base plate is provided at the lower end of the upper base plate;

[0009] Step S2: The host computer sends a start signal to the microphone array. After receiving the response signal, the host computer enters the ready state, and the microphone array under test enters the real-time horn detection state.

[0010] Step S3: The host computer controls the position and speed of the speaker through pulses according to the initialization parameters, and the speaker and noise source continuously play audio;

[0011] Step S4: The horn's horn-firing model outputs two probabilities, q1 and q2, respectively; if the probability of the horn sound is higher, the horn-firing localization procedure is initiated to estimate the horn's position, obtaining the estimated spatial coordinates l = (l x ,l y ,l z The estimated path difference d between the microphone and the microphone kl l and k represent the microphone numbers, respectively;

[0012] Step S5: The host computer calculates the spatial rectangular coordinates s = (s...) of the speaker's position using the stepper motor's position sensor. x ,s y ,s z Meanwhile, store the probabilities p1 and p2 of the type of sound played by the speaker. If the sound played is a horn, then p1 = 0 and p2 = 1; if the sound played by the speaker is not a horn, then p1 = 1 and p2 = 2.

[0013] Step S6: The microphone array sends the calculation results to the host computer. The transmitted data includes the horn recognition probabilities q1 and q2 from step S4, the estimated spatial coordinates l of the horn, and the estimated sound path difference d. kl ;

[0014] Step S7: Repeat steps S2-S6, with the speaker position changing according to the trajectory generated in step S1, for a total of N times; save the data obtained in step S5 in the host computer; after N tests, the host computer enters the error evaluation and correction status of the microphone array;

[0015] The error assessment of the microphone array is divided into two parts: positioning accuracy and horn accuracy; the estimated spatial coordinates p and actual spatial coordinates l of the N sound sources saved in step S6 are retrieved; assuming the data of the i-th test is l iand tm i The host computer calculates the error of each measurement according to formula (1), and calculates the mean of the positioning error of N tests according to formulas (2) and (3) respectively. And variance δ;

[0016] E i =|l i -s i |,i=1,2,…,N (1)

[0017]

[0018]

[0019] Step S8: The host computer reads all saved q1, q2, p1, and p2. Assume the value of the i-th test is q. i1 q i2 p i1 and p i2 ; Calculate the horn recognition error acc of the i-th test according to formula (4); Calculate the overall error of the N tests according to formula (5).

[0020] acc i =-p i1 log2q i1 -p i2 log2q i2 ,i=1,2,...,N (4)

[0021]

[0022] Step S9: Assuming the microphone array has a total of Z microphones, the operator inputs the spatial three-dimensional coordinates M of each microphone in the array relative to microphone number 1. i (x, y, z), where the actual microphone position has an offset Δ in spatial coordinates. i (x,y,z), the actual three-dimensional coordinates TM of each microphone i =M i +Δ i Microphone No. 1 is selected as the reference microphone, and its relative spatial coordinates are constant at (0,0,0); Z is a natural number greater than 1.

[0023] Step S10: Taking the k-th microphone as an example, k≠1, and k=2,3,...Z, the host computer reads the d saved from each test. 1k This represents the path difference between microphone k and microphone 1; assuming the path difference between microphone k and microphone 1 in the i-th iteration is d. 1kiThen it satisfies the geometric relationship of formula (6); microphone 1 is the reference origin, satisfying |M1+Δ1-s i |=|M1+s i Substituting | and simplifying by squaring, we get formula (7); assuming e = |Δ k Substitute into formula (7); let the microphone deviation Δb = [Δ; e 2 Using N sets of data matrix formula (7), we obtain formula (8), and then use the overdetermined linear equation system solution formula to obtain Δb. The first three terms are the Δb to be solved. k ;

[0024] |M K +Δ K -S|-|M1+Δ1-s i |=d 1ki (6)

[0025]

[0026] in

[0027]

[0028] Step S11: Repeat step S10 on the host computer Z-1 times to calculate the positional deviation Δ of all microphones in the microphone array, and then save the calculation results.

[0029] To further achieve the purpose of this invention, preferably, the determination of the total number of tests N is determined by the purpose of the tests.

[0030] Preferably, the mobile platform is equipped with a synchronous pulley, which is divided into two groups according to different heights and arranged in a CoreXY structure.

[0031] Preferably, the lower base plate is equipped with two stepper motors at both ends, and the other end of the lower base plate and the upper base plate are equipped with eight synchronous pulleys of different heights; there are four openings below the connector, and the two ends of the two synchronous belts are fixed to the openings of the connector.

[0032] Preferably, the horn is fixed to the connector by screws.

[0033] Preferably, in step S4, the speaker plays an audio clip and then stops playing.

[0034] Preferably, the mean The smaller the value of the variance δ, the better the positioning effect of the microphone array. The host computer will calculate the mean. And the variance δ is preserved.

[0035] Preferably, in step S8, the horn recognition error acc of the i-th test is calculated according to formula (4); if the model recognition is accurate, acc i The value approaches 0, otherwise it approaches 1.

[0036] Preferably, in step S8, the overall error of the N tests is calculated according to formula (5). The closer the value is to 0, the better the microphone array recognizes the horn. The host computer will... save.

[0037] Preferably, the host computer outputs the data saved in steps S7-S8 and S11 via command line. δ、 And Δ.

[0038] Compared with the prior art, the advantages of the present invention are as follows:

[0039] 1) This invention simulates the movement of a sound source on the road surface by using a motor to drive the horn, making the test scenario closer to the actual working scenario; the horn can be replaced with the desired sound source to test the localization effect of different types of sound sources, and the sound source from the horn is an actual broadband sound source, while the microphone array cannot only work under the narrowband assumption; and its position can be read directly, while the test method of changing the position of an actual stationary sound source multiple times will introduce measurement errors.

[0040] 2) The device under test in this invention is connected to the host computer only through a cable, which is convenient for disassembly. To test different devices, you only need to save the solution location information and timestamp as required to generate relevant information of the device under test.

[0041] 3) Actual road tests cannot obtain the true value of the estimated position. Using this device to measure the positioning performance of the microphone array can yield more accurate results.

[0042] 4) This invention can correct the microphone array while measuring its performance, providing a reference for equipment optimization.

[0043] 5) This invention can simulate various scenarios, making actual testing of microphone arrays more convenient. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of a device for testing the performance of a microphone array according to the present invention.

[0045] Figure 2 for Figure 1 A schematic diagram of the China Mobile platform.

[0046] Figure 3 This is a schematic diagram of the transmission system structure of the present invention.

[0047] Figure 4This is a schematic diagram illustrating the direct propagation of sound according to the present invention.

[0048] Figure 5 This is a schematic diagram illustrating the simulated NLOS conditional propagation of sound according to the present invention.

[0049] Figure 6 This is a comparison chart of the test results for the correction method of the present invention.

[0050] Figure 7 This is a flowchart of the testing process for this invention.

[0051] The annotations in the attached figures are explained as follows:

[0052] 1. Host computer; 2. Cable; 3. Microphone array; 4. Moving platform; 5. Stepper motor; 6. Guide rail; 7. Synchronous pulley; 8. Limit switch; 9. Speaker; 10. Slider; 11. Upper base plate; 12. Connector; 13. Lower base plate. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the application will be further described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0054] This invention simulates static, dynamic, non-line-of-sight, and noisy environments by building a platform to test the performance of microphone arrays, mainly measuring array recognition rate and array positioning accuracy. The microphone array consists of multiple microphones, and the installation error of the microphones will also have a significant impact on the positioning effect. The method of this invention can measure the installation error of each microphone array element.

[0055] For positioning performance testing, the tested device outputs estimated coordinates under different simulated environments, and the device of this invention outputs reading coordinates. After multiple measurements, the error is calculated. Car horn recognition belongs to sound recognition, and the probability of the corresponding label is output each time. The horn recognition error is calculated by formulas (4)-(5). When the output probability is closer to the real label, the cross-entropy result is closer to 0, and vice versa.

[0056] In a microphone array, each microphone has a pre-defined installation position, but prolonged storage, collisions, or vibrations can cause changes in the installation angle and position. In addition to measuring the horn recognition and horn positioning performance of the device under test, this invention also provides a method for correcting the position of the microphone array by calculating the deviation of the spatial coordinates to compensate for existing coordinates. Specifically, steps S9-S11 are shown.

[0057] The horn of this invention is fixed to the connector by studs. By changing the type of microphone, such as a basin-shaped horn, a snail-shaped horn, and a cylindrical air horn, the positioning effect of a horn device on different types of horns can be simulated. In addition, the number of installed horns can also be changed to simulate monotone and multitone horns.

[0058] In addition to the sound source, the testing equipment of this invention can place interference microphones around the device to simulate background noise. The simulated noise typically includes ambient sound, engine noise, and other horn sounds, such as those from electric bicycles. It can also simulate the performance of the microphone array under different signal-to-noise ratios depending on the volume of the played sound. The speed of the testing system can be controlled by pulse signals and is variable in real time, and the trajectory of the speaker is controllable.

[0059] The test equipment of this invention simulates motion with a maximum speed of 30 km / h. According to Equation 1-1, the angular velocity of the synchronous pulley can be obtained. The synchronous pulley diameter is 81.49 mm and the pitch is 8 mm. According to the mechanical design manual, the synchronous pulley transmission efficiency is 98%. Under the two-stage transmission condition, the maximum angular velocity of the servo motor should be greater than 213 rad / s and the maximum speed should be greater than 2034 r / min. Considering the small motor load, the motor model SVGA-10B21AD is selected, and the parameters are shown in Table 1.

[0060] v=w*r (1-1)

[0061] Where w is the angular velocity, v is the linear velocity, and r is the design radius of the synchronous pulley.

[0062] Table 1. Motor Parameters

[0063]

[0064] The host computer on the test equipment determines the simulation mode and the number of simulations, and initializes the entire system.

[0065] The initialization count is sent to the device under test, and a start command is sent once. After receiving the start command, the device under test sends an ACK response command to the test device.

[0066] The test device moves to the specified (X,Y,Z) coordinates, generating a horn and noise. After sending an ACK signal, the device under test enters a waiting state to determine if a horn has been generated. If a horn is detected, it performs localization, sets a timeout internally, and a timeout indicates that no horn was detected. If a horn is detected, the path difference between different microphones and the sound source is calculated, and the spatial three-dimensional coordinates of the sound source are estimated. The microphone array periodically sends the calculation results to the host computer.

[0067] After sounding the horn for a certain period of time, the testing device stops playing the horn and waits to receive the calculation results from the device under test. After receiving the data, it saves the data.

[0068] After the entire test is completed, the host computer reads the previously saved data and calculates the horn positioning error, horn recognition error, and microphone array correction coefficient of the device under test. Simultaneously, the signal-to-noise ratio and propagation conditions can be modified for multiple tests to obtain the overall performance of the device under test. The results are saved and displayed via command line.

[0069] like Figure 1 As shown, a schematic diagram of a testing system includes a host computer 1, which is installed next to a mobile platform 4 and connected to the mobile platform 4 and the microphone array 3 under test via cables 2. The host computer 1 controls the forward and reverse rotation of a stepper motor via cables, obtains the motor rotation angle from the step angle and the number of pulses, and obtains the position of the speaker 9 in the X and Y axis directions. The host computer 1 obtains the orientation calculation results of the microphone array 3 in real time and saves the read results in a specified format.

[0070] like Figure 2 As shown, the mobile platform 4 includes a stepper motor 5, guide rails 6, synchronous pulleys 7, limit switches 8, a horn 9, a slider 10, an upper base plate 11, a connecting piece 12, and a lower base plate 13. The horn 9 is mounted on the connecting piece 12, and both ends of the connecting piece 12 are movably connected to two parallel guide rails 6 via sliders 10. The connecting piece 12 is connected to the stepper motor 5 via a synchronous belt. A limit switch 8 is installed at the end of the guide rails 6. The guide rails 6 are mounted on the upper base plate 11, and a lower base plate 13 is provided at the lower end of the upper base plate 11. Preferably, the lower base plate 13 is equipped with a limit switch 8 at both ends. Two stepper motors 5 are installed on the other end of the lower base plate 13 and the upper base plate 11, and eight synchronous pulleys 7 of different heights are installed on them. There are four openings below the connector 12, and the two ends of the two synchronous belts are fixed to the openings of the connector 12. When the stepper motors 5 rotate, they generate a pulling force on the connector 12. The slider 10 is installed on the guide rail 6 and does not provide constraints in the X-axis and Y-axis directions. The connector 12 and the horn 9 are pulled by the stepper motors and generate displacement in the X-axis and Y-axis directions. When the connector 12 or the upper base plate 11 touches the limit switch during movement, it will stop moving forward.

[0071] The speaker 9 is preferably fixed to the connector 12 with screws. During the test, the speaker 9 can be disassembled and replaced with other types of sound sources.

[0072] like Figure 3 As shown, the mobile platform 4 is equipped with 8 synchronous pulleys 5, which are divided into two groups according to different heights and arranged in a CoreXY structure to change the direction of synchronous belt tension; the two stepper motors 5 rotate in opposite directions, and the upper base plate 11, connecting piece 12 and horn 9 move along the Y-axis; the two stepper motors 5 rotate in the same direction, and the connecting piece 12 and horn 9 move along the X-axis.

[0073] like Figure 4As shown, the simulation shows direct sound propagation. By modifying the controllable noise source, different road surface noises can be simulated, including ordinary background noise or sudden collision sounds. The array performance can then be calculated. The sound source S is driven by a servo motor to change its (X, Y) coordinates, which can simulate the horn sound of a vehicle. The speed is controllable and observable, which can effectively test the performance of the array signal under the influence of the Doppler effect.

[0074] like Figure 7 As shown, a method for testing the performance of a microphone array includes the following steps:

[0075] Step S1: Start the host computer 1 and the microphone array under test 3. The host computer 1 determines the trajectory of the sound source through the test purpose and manual operation, and obtains the total number of tests N; N is a natural number greater than 1.

[0076] Step S2: The host computer 1 sends a start signal to the microphone array 3 and waits for a response signal from the array 3. After receiving the response signal, the host computer enters the ready state. At this time, the microphone array 3 under test also enters the ready state and enters the real-time horn detection state.

[0077] Step S3: After the host computer 1 enters the ready state, it controls the position and speed of the speaker 9 through pulse control according to the initialization parameters, so that the speaker 9 is stationary or moving on the moving platform 4. The speaker 9 and the noise source start playing audio for a period of time.

[0078] Step S4: Microphone array 3 continuously reads data for horn detection. Because it is a binary classification problem, the horn model outputs probabilities q1 and q2 for two types of horn sounds. If the probability of the horn sound is high, the horn localization procedure is entered to estimate the position of horn 9, obtaining the estimated spatial coordinates l = (l x ,l y ,l z The estimated path difference d between the microphone and the microphone kl L and K represent the microphone numbers, respectively; speaker 9 plays an audio clip and then stops playing.

[0079] Step S5: The host computer 1 calculates the spatial rectangular coordinates s = (s...) of the speaker's position using the position sensor of the stepper motor. x ,s y ,s z Meanwhile, based on the type of sound played by speaker 9, save p1 and p2. If the sound played is a horn, then p1 = 0 and p2 = 1; if speaker 9 plays a sound other than a horn, then p1 = 1 and p2 = 2.

[0080] Step S6: The host computer 1 waits to receive the results from the array; the microphone array 3 sends the calculation results to the host computer 1. The transmitted data includes the following parts: the horn recognition results q1 and q2 from step S4, the estimated spatial coordinates l of the horn 9, and the estimated sound path difference d. kl The host computer 1 saves the received data.

[0081] Step S7: Repeat steps S2-S6, with the sound source position changing according to the trajectory generated in S1, for a total of N times; save the data obtained in step S5 in the host computer 1; after N tests, the host computer 1 enters the error evaluation and correction status of the microphone array 3;

[0082] The error assessment of microphone array 3 is divided into two parts: positioning accuracy and horn accuracy; the N estimated sound source spatial coordinates p and the actual spatial coordinates tm saved in step S6 are retrieved; assuming the data of the i-th test is l i and tm i The host computer 1 calculates the error of each measurement according to formula (1), and continues to calculate the mean of the positioning error of N tests according to formulas (2) and (3). Sum of variance δ, mean The smaller the value of the variance δ, the better the positioning effect of microphone array 3. Finally, the mean value is calculated. And the variance δ is preserved.

[0083] E i =|l i -s i |,i=1,2,…,N (1)

[0084]

[0085]

[0086] Step S8: The host computer reads all saved q1, q2, p1, and p2. Assume the value of the i-th test is q. i1 q i2 p i1 and p i2 ; Calculate the horn recognition error acc for the i-th test according to formula (4); if the model recognizes accurately, acc i The value approaches 0, otherwise it approaches 1; then calculate the overall error of N tests according to formula (5). The closer the value is to 0, the better the horn recognition effect of array 3. The host computer 1 will... save.

[0087] acc i =-p i1 log2q i1 -pi2 log2q i2 ,i=1,2,...,N (4)

[0088]

[0089] Step S9: Assuming that microphone array 3 has a total of Z microphones, the operator inputs the spatial three-dimensional coordinates M of each microphone in array 3 relative to microphone number 1. i (x, y, z), due to various factors, the actual position of each microphone in the spatial coordinate system has an offset Δ. i (x,y,z), therefore the actual three-dimensional coordinates TM of each microphone i =M i +Δ i Microphone number 1 is selected as the reference microphone, and its relative spatial coordinates are constant at (0,0,0). Z is a natural number greater than 1.

[0090] Step S10: Taking microphone k as an example, k≠1, and k=2,3,..Z, host computer 1 reads the d saved from each test. 1k This represents the path difference between microphone k and microphone 1; assuming the path difference between microphone k and microphone 1 in the i-th iteration is d. 1ki Then it satisfies the geometric relationship of formula (6). Since microphone 1 is the reference origin, |M1+Δ1-s i |=|M1+s i Substituting | and simplifying by squaring, we get formula (7); since formula (7) cannot be solved directly, we assume e = |Δ k Substitute into formula (7); let the microphone deviation Δb = [Δ; e 2 Using N sets of data matrix formula (7), we obtain formula (8), and then use the overdetermined linear equation system solution formula to obtain Δb. The first three terms are the Δb to be solved. k .

[0091] |M K +Δ K -S|-|M1+Δ1-s i |=d 1ki (6)

[0092]

[0093] in

[0094]

[0095] Step S11: Repeat step S10 Z-1 times on the host computer 1 to calculate the positional deviation Δ of all microphones in microphone array 3, and then save the calculation results. Microphone array 3 uses the saved Δ to reduce the solution error.

[0096] Step S12: The host computer 1 outputs the data saved in steps 7-8 and S11 via the command line. δ、 And Δ.

[0097] like Figure 5 As shown, road conditions can occasionally result in NLOS (Normally Unstable) situations, which often increase the positioning algorithm error. By adding obstacles between the sound source and the microphone, this road condition can be simulated, and the array performance can be calculated.

[0098] The simulation was conducted following the steps of a method for testing microphone array performance. This simulation was designed with nine sound source locations (speaker locations), and 56 horn blasts were tested at each location for localization. Figure 6 As shown in the figure, the positioning error of microphone array 13 before correction is basically above 0.2m; after correction of microphone array 13 through the above steps S7-S11, the positioning error is significantly reduced under the condition that other conditions remain unchanged, and is below 0.2m, with the error basically in the range of 0.1m and below.

[0099] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for testing the performance of a microphone array, characterized in that... Includes the following steps: Step S1: Start the host computer and the microphone array under test, and determine the total number of tests. ; It is a natural number greater than 1; the host computer is installed next to the mobile platform and is connected to the mobile platform and the microphone array under test respectively through cables; the speaker of the mobile platform is installed on the connector, and the two ends of the connector are movably connected to two parallel guide rails through sliders; the connector is connected to the stepper motor through a synchronous belt; a limit switch is installed at the end of the guide rail; the guide rail is set on the upper base plate, and a lower base plate is provided at the lower end of the upper base plate; Step S2: The host computer sends a start signal to the microphone array under test. After receiving the response signal, the host computer enters the ready state, and the microphone array under test enters the real-time horn detection state. Step S3: The host computer controls the position and speed of the speaker through pulses according to the initialization parameters, and the speaker and noise source continuously play audio; Step S4: The horn-honking model outputs the probabilities of two types as follows: and If the probability of a horn sound is high, the system enters the horn location procedure to estimate the horn's position and obtain its estimated spatial coordinates. Estimated path difference between the microphone and the microphone , and These represent the microphone numbers; Step S5: The host computer calculates the spatial rectangular coordinates of the speaker's position using the position sensor of the stepper motor. At the same time, it stores the probability of the type of sound played by the speaker. and If the sound played is a horn, then ; If the loudspeaker is playing a sound other than a horn, then... ; Step S6: The microphone array under test sends the calculation results to the host computer. The transmitted data includes the horn recognition probability from step S4. and Estimated spatial coordinates of the loudspeaker and estimated sound path difference ; Step S7: Repeat steps S2-S6, with the speaker position changing according to the trajectory generated in step S1, for a total of N times; save the data obtained in step S5 in the host computer; after N tests, the host computer enters the error evaluation and correction status of the microphone array under test. The error assessment of the microphone array under test is divided into two parts: positioning accuracy and whistling accuracy; the N estimated sound source spatial coordinates saved in step S6 are... and actual spatial coordinates Extract; assuming the data from the i-th test is and The host computer calculates the error of each measurement according to formula (1), and calculates the mean of the positioning error of N tests according to formulas (2) and (3) respectively. and variance ; Step S8: The host computer reads all saved data. , , and Assume the value of the i-th test is , , and ; The horn recognition error for the i-th test is calculated according to formula (4). ; Calculate the overall error of N tests according to formula (5) ; Step S9: Assuming the microphone array under test has a total of Z microphones, the operator inputs the spatial three-dimensional coordinates of each microphone in the array relative to microphone number 1. In reality, the position of each microphone is offset in spatial coordinates. The actual three-dimensional coordinates of each microphone ; Microphone No. 1 is selected as the reference microphone, and its relative spatial coordinates are constant at (0,0,0); Z is a natural number greater than 1; i Step S10: Taking microphone k as an example, where k≠1 and k=2,3,...Z, the host computer reads the data saved from each test. This represents the path difference between microphone k and microphone 1; assuming the path difference between microphone k and microphone 1 in the i-th iteration is... If , then it satisfies the geometric relationship of formula (6); Microphone #1 is the reference origin, satisfying... Substituting the square and simplifying, we get formula (7); assuming Substitute into formula (7); let the microphone deviation to be solved be... Formula (8) is obtained by using N sets of data matrix transformation formula (7), and then solved using the overdetermined linear equation system solution formula. The first three terms are the ones that need to be solved. ; in ; Step S11: Repeat step S10 on the host computer Z-1 times to calculate the positional deviation of all microphones in the microphone array under test. Then save the calculation results.

2. The method for testing microphone array performance according to claim 1, characterized in that, The determination of the total number of tests It is determined by the purpose of the test.

3. The method for testing microphone array performance according to claim 1, characterized in that, The mobile platform is equipped with a synchronous pulley, which is divided into two groups based on different heights and arranged in a CoreXY structure.

4. The method for testing microphone array performance according to claim 3, characterized in that, The lower base plate is equipped with two stepper motors at both ends, and eight synchronous pulleys of different heights are installed at the other end of the lower base plate and the upper base plate; there are four openings below the connector, and the two ends of the two synchronous belts are fixed to the openings of the connector.

5. The method for testing the performance of a microphone array according to claim 1, characterized in that, The speaker is fixed to the connector with screws.

6. The method for testing the performance of a microphone array according to claim 1, characterized in that, Step S4: The speaker plays an audio clip and then stops playing.

7. The method for testing the performance of a microphone array according to claim 1, characterized in that, The mean and variance The smaller the value, the better the positioning effect of the microphone array under test. The host computer will calculate the mean. and variance save.

8. The method for testing the performance of a microphone array according to claim 1, characterized in that, In step S8, the horn recognition error of the i-th test is calculated according to formula (4). ; If the model recognizes accurately The value approaches 0, otherwise it approaches 1.

9. The method for testing the performance of a microphone array according to claim 1, characterized in that, In step S8, the overall error of the N tests is calculated according to formula (5). , The closer the value is to 0, the better the microphone array under test recognizes the horn. The host computer will... save.

10. The method for testing the performance of a microphone array according to claim 1, characterized in that, The host computer outputs the data saved in steps S7-S8 and S11 via command line. , , and .

Images