Robust sound source localization method fusing sound intensity amplitude inter-frame correlation

Abstract

The invention discloses a robust sound source localization method fusing sound intensity amplitude inter-frame correlation, and the method comprises the steps: receiving a time domain voice signal through a regular triangle array, and carrying out the short-time Fourier transform of the time domain signal, and obtaining a time-frequency domain signal; calculating a sound intensity component by using a sound intensity estimation method of sound pressure Taylor series expansion; calculating a comprehensive weight of each time frequency point by using a sound intensity estimation consistency algorithm, and calculating a sound intensity amplitude at the same time; weighting the sound intensity amplitude by using the comprehensive weight; constructing a reliable time-frequency point screening factor fusing the inter-frame correlation of the sound intensity amplitudes by utilizing the strong correlation between the weighted sound intensity amplitudes and the unweighted sound intensity amplitudes; screening out time-frequency points which are less influenced by reverberation and noise by using the screening factor, and superposing the sound intensity corresponding to the screened time-frequency points to obtain an azimuth angle estimated value; the method provided by the invention has better anti-reverberation and anti-noise capabilities in indoor sound source localization.

Description

Technical Field

[0001] The invention relates to a robust sound source localization method integrating inter-frame correlation of sound intensity amplitude, and belongs to the technical field of indoor sound source localization. Background Art

[0002] Sound intensity refers to the average sound energy flow per unit area perpendicular to the direction of sound propagation. It is a vector closely related to the direction of arrival (DOA) of the sound source. The sound intensity vector is estimated through an algorithm to estimate the azimuth of the sound source. The equilateral triangle array is a non-orthogonal first-order difference array. If we refer to the method of estimating sound intensity using differential approximation in the orthogonal first-order difference array of four microphones, we first need to estimate the sound pressure at the midpoint of the array based on the average sound pressure of the three microphones, and then project the velocity components estimated by the three pairs of dipoles of the shared microphone onto the coordinate axes and superimpose them. This method has the problems of large computational complexity and inaccurate sound intensity estimation, which leads to poor sound source localization effect.

[0003] After estimating sound intensity, not all time-frequency points correspond to sound intensities that are useful for sound source localization. In fact, only a small number of time-frequency points are reliable and useful for sound source localization. Therefore, selecting reliable time-frequency points that are useful for sound source localization is key to improving the algorithm's ability to resist reverberation and noise. The pseudo-intensity estimation consistency (EC) algorithm uses the high consistency between pseudo-intensities (PIVs) in time-frequency regions dominated by the direct path of a single sound source to select time-frequency points. It offers fast computational speed, good reverberation resistance, and high versatility, making it applicable to sound intensity estimation algorithms. However, it also has drawbacks: the normalized sound intensity is used to calculate time frame weights, resulting in sound intensity vectors with different amplitudes contributing equally to the time frame weight, ignoring the sound intensity amplitude information of individual time-frequency points. Furthermore, the EC algorithm only considers information from the current frame when calculating the weights for individual frequency points, ignoring information from preceding and following frames. This can improve the accuracy of direct sound detection. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a robust sound source localization method that integrates the inter-frame correlation of sound intensity amplitude. Based on the estimated sound intensity, a time-frequency point screening method that integrates the inter-frame correlation of sound intensity amplitude is adopted to effectively screen out time-frequency points that are less affected by reverberation and noise, thereby improving the anti-reverberation and anti-noise capabilities of the method in indoor sound source localization scenarios.

[0005] The present invention adopts the following technical solutions to solve the above technical problems: A robust sound source localization method integrating inter-frame correlation of sound intensity amplitude comprises the following steps: Step 1: Use an equilateral triangle array consisting of three omnidirectional microphones to receive the sound source signal, and convert the time-domain sound pressure signal received by the microphones into the time-frequency domain through short-time Fourier transform. In the time-frequency domain, the sound intensity estimation method of the sound pressure Taylor series expansion is used to calculate the sound intensity component of each time-frequency point at the center of the equilateral triangle array; Step 2: Calculate the sound intensity amplitude based on the sound intensity components calculated in step 1, calculate the comprehensive weight corresponding to each time-frequency point using the sound intensity estimation consistency method, and weight the sound intensity amplitude using the comprehensive weight to obtain the weighted sound intensity amplitude; Step 3: construct a time-frequency point screening factor based on the inter-frame correlation of the sound intensity amplitude according to the weighted sound intensity amplitude and the unweighted sound intensity amplitude, and screen out reliable time-frequency points from all time-frequency points according to the time-frequency point screening factor; Step 4: Superimpose the sound intensity components corresponding to the reliable time-frequency points screened out in step 3, and obtain the estimated value of the sound source azimuth angle through the sound intensity localization method.

[0006] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects: 1. The present invention uses an equilateral triangle array to estimate sound intensity and then localize the sound source. Compared with a four-microphone uniform circular array, it has lower computational complexity and has the same azimuth omnidirectional spatial positioning capability as a four-microphone uniform circular array.

[0007] 2. The present invention extends the sound intensity estimation method of Taylor series expansion in three-dimensional regular tetrahedron arrays to two-dimensional planar arrays. Compared with methods such as orthogonal decomposition of sound pressure difference and cross-power spectrum, this method is more convenient to calculate and has comparable positioning performance.

[0008] 3. Building on the sound intensity estimation consistency detection method, this invention fully utilizes the sound intensity amplitude information of individual time-frequency points. Using the consistency detection method, it calculates a comprehensive weight to weight the sound intensity amplitude, and then utilizes the inter-frame correlation between the two. This time-frequency point screening method, incorporating the inter-frame correlation of sound intensity amplitude, can more accurately detect time-frequency points that are less affected by reverberation and noise, thereby improving the positioning accuracy of the present method.

[0009] 4. The present invention has high positioning accuracy in indoor sound source positioning in simulation scenes, is relatively robust to the interference of noise and reverberation, and has certain practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is an equilateral triangle array of three microphones designed by the present invention; Figure 2 This is a schematic diagram of the indoor sound source localization simulation scene; Figure 3(a) and Figure 3(b) are the comparisons of the root mean square error of the method of the present invention with the SI, SI-EC, and DPD-MUSIC algorithms under different reverberation time conditions when the signal-to-noise ratio is fixed at 15dB and 5dB, respectively; FIG4(a) and FIG4(b) are respectively comparisons of the root mean square error of the method of the present invention and the SI, SI-EC, and DPD-MUSIC algorithms under different signal-to-noise ratio conditions when the reverberation time is fixed at 300ms and 800ms; Figure 5 This is a comparison of the root mean square error of the method of the present invention with the SI, SI-EC, and DPD-MUSIC algorithms under different array radius conditions when the reverberation time is 0.5s and the signal-to-noise ratio is 10dB; Figure 6 This is a flow chart of a robust sound source localization method integrating inter-frame correlation of sound intensity amplitudes according to the present invention. DETAILED DESCRIPTION

[0011] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be interpreted as limiting the present invention.

[0012] The present invention proposes a robust sound source localization method integrating inter-frame correlation of sound intensity amplitude. Figure 6 It is a flow chart of the method of the present invention, and its specific implementation steps are as follows: Step 1, consider three omnidirectional microphones 、 、 By radius , equal azimuth interval Composition Figure 1 The equilateral triangle array shown. Then the microphone receives the time domain signal Perform short-time Fourier transform and estimate the sound intensity component in the time-frequency domain using the Taylor series expansion method of sound pressure and .

[0013] 1.1, for the indoor single sound source positioning scenario, the far-field plane wave sound source Incident direction and coordinate system The angle between the positive semi-axis and the azimuth is defined as , its value range is , the time domain sound pressure signals received by the three microphones It can be expressed as: (1) in, , is a pure speech signal in the time domain, symbol represents the convolution operation, Indicates the sound source to The room impulse response of the microphones For the The additive noise signals of the microphones are assumed to be independent of each other and uncorrelated with the sound source signal.

[0014] 1.2. Since speech signals have short-term stationary characteristics, the time domain observation signal is generally subjected to short-time Fourier transform (STFT), and the above formula is expressed as: (2) in, and Respectively represent time units and frequency units, 、 、 as well as They are 、 、 and The signal after STFT.

[0015] 1.3. In the time-frequency domain, the sound intensity estimation algorithm based on the Taylor series expansion of the sound pressure in the three-dimensional regular tetrahedron array is extended to the two-dimensional regular triangle array. The sound pressure and the absolute coordinates of the microphone are input and then the sound pressure and vibration velocity are calculated through matrix transformation. The frequency is omitted for the convenience of derivation. , the specific derivation process is as follows: In an ideal streaming medium, the relationship between particle velocity and acoustic pressure gradient is given by the linear Euler equation: (3) in, is the air density, and The particles are The vibration velocity and sound pressure at each moment, Represents a gradient operation.

[0016] The particle vibration velocity is expressed by the acoustic pressure gradient as: (4) in, is the angular frequency. The finite difference approximation is usually used to approximate one component of the acoustic pressure gradient, and then Equation (4) is used to calculate the particle velocity.

[0017] The sound pressure at the midpoint of two closely spaced microphones can be obtained using the zero-order and first-order terms of the Taylor series expansion, which are omitted for ease of derivation. and , then it is located at the origin of the coordinate system Position Microphone sound pressure Expressed as: (5) in, is the sound pressure at the origin, , It is The sound pressure gradient at each microphone relative to the origin.

[0018] Taking the regular triangle array as an example, the three microphones correspond to the position coordinates 、 、 Substituting into equation (5), we can write it in the form of the following three linear equations: (6) in, , , in order to facilitate subsequent derivation, the above formula is rewritten into matrix form.

[0019] make , , Figure 1 Microphone relative position coordinate matrix Expressed as: (7) Pair Matrix Seek the inverse and then Multiply them to get the vector related to sound pressure and vibration velocity for: (8) in, 、 、 It is omitted , corresponding to formula (2) ; by vector The sound pressure at the center point can be approximated as: , The directional sound pressure gradient is , The directional sound pressure gradient is .

[0020] 1.4, according to the corresponding 、 、 , estimate the array center point Direction and The sound intensity components in the direction are: (9) (10) in, is the imaginary unit, This is the real part operation.

[0021] Step 2: Calculate the sound intensity amplitude based on the sound intensity component estimated in step 1 , and then use the sound intensity estimation consistency method to calculate the comprehensive weight , use the comprehensive weight to weight the sound intensity amplitude, and get the weighted sound intensity amplitude .

[0022] The traditional DOA estimation algorithm based on sound intensity is to calculate the DOA of all time-frequency points. and The components are superimposed separately to achieve the effect of anti-noise and anti-reverberation. Therefore, according to equations (9) and (10), the closed-form solution based on the azimuth angle estimation value of the regular triangle array is: (11) As shown in formula (11), most positioning methods based on sound intensity estimation obtain a positioning result by superimposing the sound intensity components of all time-frequency points. This is a positioning method based on sound intensity estimation without time-frequency point screening. The following is a positioning method based on time-frequency point screening based on sound intensity inter-frame correlation derived and proposed by the present invention.

[0023] When a certain time-frequency point is dominated by a single sound source direct path, its sound intensity vector can be expressed as: (12) in, For pure speech signal After STFT transformation, the signal is the azimuth of the direct sound, represents a direct path, is transposed. Direction and The directional sound intensity is expressed as: , .

[0024] When the direct sound is dominant, the sound intensity amplitude is: (13) in, is the vector norm. When a frame signal is dominated by direct sound, the sound intensity of most time-frequency points in this frame is concentrated in the direction of the sound source, which is called a concentrated frame.

[0025] 2.1. Calculate the normalized weight of each frame based on the correlation between the sound intensity of a single time-frequency point and the sound intensity of all time-frequency points in a frame. , specifically: (14) Where, is the number of time-frequency points in a frame, Represents the sound intensity vector of a single time-frequency point.

[0026] When a frame is an aggregated frame, assuming that all time-frequency points in this frame are dominated by direct sound, substituting equations (12) and (13) into (14), we can see that when the first When the frame is dominated by direct sound, , indicating that most of the sound intensity vectors in this frame are concentrated in a specific direction. When , it indicates that due to the existence of reverberation and noise, the sound intensity vector in this frame points randomly.

[0027] 2.2. In fact, in each frame signal, only some time-frequency points may be dominated by direct sound. The weight of each time-frequency point can be calculated by the spatial distance between the sound intensity vector of each time-frequency point and the average sound intensity vector in a frame. , as follows: (15) in, Indicates the The sound intensity mean of the frame signal can be known from formula (15) that if the sound intensity vector of a single time-frequency point is and the average sound intensity vector in a frame When they are equal, it indicates that the spatial distance between the two is very small. According to the properties of the inverse cosine function, at this time .

[0028] 2.3, the current frame weight and the weight of each time-frequency point in a frame Multiply to get the comprehensive weight of each time-frequency point: (16) When a frame is an aggregate frame, the time frame weight , and when a certain time-frequency point in this frame is dominated by direct sound, the weight of a single time-frequency point , then from formula (16) we can know that the comprehensive weight at this time is Also close to 1.

[0029] The above sound intensity estimation consistency method can calculate the comprehensive weight of each time-frequency point.

[0030] 2.4. In the actual positioning scenario, the sound intensity vector of each time-frequency point is calculated according to the common point sound intensity estimation method, and then its sound intensity amplitude is calculated. : (17)

[0031] 2.5, calculate the comprehensive weight corresponding to each time-frequency point using the estimation consistency algorithm , and then weight the sound intensity amplitude to obtain the weighted sound intensity amplitude , specifically: (18) From (16) and (18), we can see that when a certain frequency point is dominated by direct sound, At this time, there is a strong correlation between the two. The weighted sound intensity amplitude is almost equal to the unweighted sound intensity amplitude. When reverberation and noise exist, the correlation between the two is weakened. Therefore, the correlation between the two is considered to be used for reliable time-frequency point screening.

[0032] Step 3: Construct the time-frequency point screening factor based on the inter-frame correlation of the sound intensity amplitude , the weighted sound intensity amplitude Unweighted sound intensity Input to the filter factor, with the threshold After comparison, reliable time-frequency points are selected.

[0033] 3.1. The direct sound in the speech signal reaches the array before the reverberation, so we can consider detecting the inter-frame correlation between the weighted and unweighted sound intensity amplitudes of consecutive frames. This paper considers a single-source detection method that detects the degree of correlation between consecutive time-frequency points in a frame. Referring to the method for constructing a correlation screening factor in single-source detection, the method of finding consecutive time-frequency points in a frame is rewritten as finding consecutive frames with the same frequency index. The time-frequency point screening factor is then expressed as: (19) Among them, after considering the compromise between the amount of calculation and the accuracy of screening, for A collection of 3 consecutive frames, represents the inter-frame summation operation, Indicates the correlation between the sound intensity amplitude and the weighted sound intensity amplitude in a time-frequency analysis area.

[0034] 3.2. In each simulation, the values ​​of the screening factors calculated at all time-frequency points are sorted in descending order, and the M% largest value is taken as the dynamic threshold , and then With dynamic threshold Compare them one by one, and when the frequency point is greater than the threshold, the time frequency point is regarded as a reliable time frequency point, and the sound intensity component corresponding to the reliable time frequency point is marked as and .

[0035] Step 4: Extract the sound intensity components corresponding to the reliable time-frequency points selected in step 3 and , and then superimpose them, and then estimate the closed-form solution of the sound source azimuth angle by the sound intensity localization method: (20)

[0036] The present invention will be further described below with reference to some specific examples.

[0037] Because the present invention performs reliable time-frequency point selection based on inter-frame correlation of sound intensity amplitude, it has good anti-reverberation and anti-noise capabilities. The following simulation experiments verify the positioning performance of the present invention in indoor reverberation and noisy environments. The specific scenario and process of the simulation experiments are as follows.

[0038] Simulation scenario such as Figure 2 As shown, the simulated rectangular room size is set to , the array is located at the geometric center of the room At point , the array is a regular triangle array of 3 microphones with a radius of 2.6 cm, the formation is as follows Figure 1 shown.

[0039] In the simulation experiment, the loudspeaker was placed at the same height as the microphone array and 1.5 m from the center of the array. The signal sampling rate was set to 16 kHz, and the simulated room reverberation was generated using the image method. The microphone noise was zero-mean additive white Gaussian noise and was uncorrelated with the speech signal. The speed of sound c was 340 m / s. The frame length in the short-time Fourier transform was 512, and the frame shift was 256. Figure 2 As shown in the gray circles, the azimuth angles of the sound sources are placed in increments of 10 degrees. The interval is 100, so a total of 36 sound sources are placed at the test positions. At each sound source placement position, 5 male voices and 5 female voices are randomly selected from the TIMIT database as the test database. After removing the silent segment from each voice, a 2-second duration is captured as the experimental sound source signal. The frequency band of the signal is set to 300-5000Hz, and each selected voice is subjected to 10 Monte Carlo experiments. Therefore, a total of 100 Monte Carlo experiments are conducted at each sound source position, and a total of 3600 experiments are conducted at the 36 candidate positions. In the DPD-MUSIC algorithm, the time smoothing factor and the frequency smoothing factor are set to 3 and 4, respectively. The method of the present invention and the SI-EC algorithm extract the frequency points with the top 35% of the values.

[0040] In order to evaluate the positioning performance of each algorithm, the root mean square error is used as the evaluation index, which is defined as: (twenty one) In formula (21), Indicates the sound source placement index, Indicates When the sound source is located, The azimuth angle estimated by the simulation algorithm is express The actual azimuth of the sound source. In order to fully demonstrate the performance of various algorithms in locating different sound source positions in the two-dimensional plane with an azimuth angle of 0 to 360 degrees, the sound source position in the simulation experiment starts from 0 degrees and increases by 10 degrees, with a total of 36 candidate sound source positions. The value is 36. Indicates the number of simulation positioning times at the current sound source position. 10 voices are selected for the experiment at a sound source position, and each voice is subjected to 10 Monte Carlo times. Therefore, during the simulation experiment, The values ​​of are all 100.

[0041] To illustrate the advantages of the method of the present invention over other indoor sound source localization methods in terms of anti-reverberation and anti-noise, the following positioning performance comparison is performed between the method of the present invention and the sound intensity estimation localization algorithm (SI) without time-frequency point screening, the consistency detection algorithm based on SI (SI-EC), and the direct path detection in the modal domain (DPD-MUSIC) algorithm.

[0042] Figure 3(a) and Figure 3(b) show the reverberation time when the signal-to-noise ratio is fixed at 15dB and 5dB respectively. The root mean square error (RMSE) changes of each sound source localization algorithm at .

[0043] Analysis shows that the SI algorithm has the worst positioning performance, because it resists the influence of reverberation and noise by superimposing all time-frequency points. Many time-frequency points dominated by reverberation will affect its positioning performance. The DPD-MUSIC algorithm in the modal domain has modal aliasing when converting the speech signal from the time-frequency domain to the modal domain. The fewer the number of microphones, the greater the impact. Therefore, in the case of a three-microphone array, the positioning performance of the modal domain algorithm is poor. Since the SI-EC algorithm only uses the current frame information, it is easy to miss the starting end when detecting the starting end of the speech, which makes its anti-reverberation ability limited. The algorithm of the present invention makes full use of the inter-frame information of the sound intensity amplitude, and detects the starting end of the speech more accurately than the SI-EC algorithm, and retains the speech-dominated segment more accurately than the EC algorithm. Therefore, the accurate detection of the initial direct sound makes the algorithm of the present invention have a strong anti-reverberation ability. As shown in Figure 3(b), under strong noise conditions with a signal-to-noise ratio of 5 dB, the proposed algorithm reduces the RMSE by approximately 3 degrees compared to the SI algorithm at a reverberation time of 0.7 seconds, and improves positioning accuracy by approximately 1.3 degrees compared to the SI-EC algorithm. Overall, the proposed algorithm demonstrates very robust anti-reverberation capabilities.

[0044] Figure 4(a) and Figure 4(b) show that when the reverberation time is fixed at 300ms and 800ms respectively, the signal-to-noise ratio is set to The RMSE changes of each sound source localization algorithm at different times.

[0045] Figure 4(a) shows that when the reverberation time is 0.3s and the signal-to-noise ratio ranges from 0dB to 25dB, the root mean square error of the four algorithms is between 2 and 3 degrees, indicating that all algorithms are relatively robust to noise under low reverberation conditions. Figure 4(b) shows that the SI algorithm still has the worst positioning performance. This is because it does not perform any time-frequency point filtering and simply sums the sound intensity components of all frequency points to estimate an angle. The DPD-MUSIC algorithm still has poor noise immunity under this formation due to modal aliasing. However, SI-EC, as a pseudo-intensity filtering algorithm, has good noise immunity. The method of the present invention fully utilizes the inter-frame correlation of sound intensity amplitude to perform time-frequency point filtering, resulting in the algorithm's positioning performance having the best noise immunity under both reverberation intensities. As shown in Figure 4(a), at a low reverberation intensity of 300ms, the method of the present invention achieves a positioning accuracy improvement of nearly 1 degree compared to the EC algorithm. As shown in Figure 4(b), the greater the reverberation intensity and the lower the signal-to-noise ratio, the more obvious the improvement in positioning accuracy of the proposed method. Compared with the basic sound intensity estimation algorithm, the average RMSE under strong reverberation conditions is reduced by about 4 degrees.

[0046] Figure 5 This figure shows the positioning performance of the four algorithms under different array radius conditions when the reverberation time is 0.5s and the signal-to-noise ratio is 10dB, and the array radius increases from 0.5cm to 5cm in steps of 0.5cm.

[0047] Depend on Figure 5 It can be seen that a radius of 1 cm is the critical value that affects the positioning performance of the sound intensity method. Therefore, when the radius increases from 0.5 cm to 5 cm, the RMSE of each algorithm will tend to first decrease and then increase. After the radius exceeds 1 cm, the root mean square error of the non-orthogonal first-order difference sound intensity estimation method will show an overall upward trend as the array radius increases. The reason is that as the array radius increases, the error of the differential approximation becomes larger, which will lead to inaccurate sound intensity estimation, and ultimately cause a slight decrease in positioning performance. Therefore, the RMSE of the SI method, SI-EC method, and the method of the present invention all have a certain upward trend. Since the method of the present invention can more accurately detect the starting end of the speech through time-frequency point screening and has good anti-reverberation and anti-noise capabilities, it also shows a stable performance improvement trend under the condition of increasing array size. Compared with the basic EC algorithm, the root mean square error often maintains a 2-degree downward trend. This shows that the method of the present invention is also robust to the increase of array size in indoor sound source localization scenarios.

[0048] Based on the same inventive concept, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the aforementioned robust sound source localization method integrating inter-frame correlation of sound intensity amplitude are implemented.

[0049] Based on the same inventive concept, an embodiment of the present application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the steps of the aforementioned robust sound source localization method integrating inter-frame correlation of sound intensity amplitude are implemented.

[0050] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0051] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0052] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0053] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0054] The above embodiments are only for illustrating the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the present invention.

Claims

1. A robust sound source localization method integrating inter-frame correlation of sound intensity amplitude, characterized in that: The steps include: Step 1: Use an equilateral triangle array consisting of three omnidirectional microphones to receive the sound source signal, and convert the time-domain sound pressure signal received by the microphones into the time-frequency domain through short-time Fourier transform. In the time-frequency domain, the sound intensity estimation method of the sound pressure Taylor series expansion is used to calculate the sound intensity component of each time-frequency point at the center of the equilateral triangle array; Step 2: Calculate the sound intensity amplitude based on the sound intensity components calculated in step 1, calculate the comprehensive weight corresponding to each time-frequency point using the sound intensity estimation consistency method, and weight the sound intensity amplitude using the comprehensive weight to obtain the weighted sound intensity amplitude; Step 3: construct a time-frequency point screening factor based on the inter-frame correlation of the sound intensity amplitude according to the weighted sound intensity amplitude and the unweighted sound intensity amplitude, and screen out reliable time-frequency points from all time-frequency points according to the time-frequency point screening factor; Step 4: Superimpose the sound intensity components corresponding to the reliable time-frequency points screened out in step 3, and obtain the estimated value of the sound source azimuth angle through the sound intensity localization method.

2. The robust sound source localization method based on inter-frame correlation of sound intensity amplitude according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 1.1, use three omnidirectional microphones according to the radius , equal azimuth interval The equilateral triangle array is used to sample the pure speech signal emitted by the sound source and obtain the microphone time-domain sound pressure signal for: ， in, , is a pure speech signal in the time domain, symbol represents the convolution operation, From the sound source to the The room impulse response of the microphones For the The additive noise signal of each microphone; Step 1.2, Perform short-time Fourier transform and get microphones at each time-frequency point The time-frequency domain sound pressure signal : ， in, Indicates the The time unit is frame, Indicates the frequency units, 、 and They are 、 and The signal after short-time Fourier transform; Step 1.3, in the time-frequency domain, generalize the sound intensity estimation algorithm of the Taylor series expansion of the sound pressure in the three-dimensional regular tetrahedron array to the two-dimensional regular triangle array, and establish a rectangular coordinate system with the center of the regular triangle array as the origin. One of the microphones is located in the coordinate system Axis positive, calculate the vector related to sound pressure and vibration velocity for: ， Depend on Get the center of the equilateral triangle array at each time-frequency point The sound pressure is , The directional sound pressure gradient is , The directional sound pressure gradient is ; Step 1.4, according to 、 and , the center of the equilateral triangle array is estimated to be Direction and The sound intensity components in the direction are: ， ， in, 、 The centers of the equilateral triangle arrays are 、 The directional sound intensity component, To take the real part operation, is the imaginary unit, is the air density.

3. The robust sound source localization method based on inter-frame correlation of sound intensity amplitude according to claim 2, characterized in that: The specific process of step 2 is as follows: Step 2.1: Calculate the normalized weight of each frame based on the correlation between the sound intensity vector of a single time-frequency point and the sound intensity vectors of all time-frequency points in a frame: ， in, For the Normalized weights of frames, is the number of time-frequency points in a frame, Represents the sound intensity vector of a single time-frequency point, is the vector two norm; Step 2.2: Calculate the weight of a single time-frequency point based on the spatial distance between the sound intensity vector of a single time-frequency point and the average sound intensity vector in a frame. : ， in, Indicates the The mean sound intensity of the frame signal; Step 2.3, normalize the weight of each frame and the weight of each time-frequency point in a frame Multiply them together to get the comprehensive weight corresponding to each time-frequency point : ， Step 2.4, the sound intensity component estimated according to step 1.4 and Calculate sound intensity amplitude : ， Step 2.5, using comprehensive weights Sound intensity amplitude Perform weighting to obtain the weighted sound intensity amplitude : 。 4. The robust sound source localization method based on inter-frame correlation of sound intensity amplitude according to claim 3, characterized in that: The specific process of step 3 is as follows: Step 3.1, the sound intensity amplitude calculated according to step 2.4 and the weighted sound intensity amplitude obtained in step 2.5 , construct the time-frequency point screening factor based on the inter-frame correlation of sound intensity amplitude : ， in, For the and A collection of three consecutive frames; Step 3.2: Sort the time-frequency point screening factors corresponding to all time-frequency points in descending order, and select the M% largest time-frequency point screening factor as the threshold , M is the preset value, and the time-frequency point screening factor corresponding to each time-frequency point is compared with the threshold By comparison, the time-frequency point screening factor is greater than The time-frequency point is the reliable time-frequency point, and the sound intensity component corresponding to the reliable time-frequency point is marked as and .

5. The robust sound source localization method based on inter-frame correlation of sound intensity amplitude according to claim 4, characterized in that: The specific process of step 4 is as follows: The sound intensity components corresponding to the reliable time-frequency points are superimposed, and the estimated value of the sound source azimuth angle in the form of a closed-form solution is obtained by the sound intensity localization method. : 。 6. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor executes the computer program, the steps of the robust sound source localization method based on inter-frame correlation of sound intensity amplitude are implemented as described in any one of claims 1 to 5.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the robust sound source localization method based on inter-frame correlation of sound intensity amplitude are implemented as claimed in any one of claims 1 to 5.

Images