Adaptive filtering algorithm and device for sound-intelligence membrane in multi-noise source environment

Abstract

The present invention provides an adaptive filtering algorithm and device for a multi-noise source environment of a sound-intelligence membrane. The algorithm analyzes the physical structure of a performance scene within the sound-intelligence membrane to determine a plurality of possible sound pickup points. The sound pickup environment corresponding to each sound pickup point within the sound-intelligence membrane is determined, and a corresponding noise source set is determined based on the sound pickup environment. The algorithm acquires and determines the corresponding noise source set for the point of the sound pickup device in real time, and determines a corresponding filtering target based on each noise source set to obtain an adaptive target after deduplication. The algorithm determines a corresponding filtering algorithm based on the adaptive target, and filters the audio of the sound pickup device. The algorithm can determine the noise sources at different positions in real time, select a suitable filtering algorithm, and improve the audio filtering effect.

Description

Technical Field

[0001] The present invention relates to data processing technology, and in particular to an adaptive filtering algorithm and device for a sound-intelligence membrane in a multi-noise source environment. Background Art

[0002] In performances in large venues, high-quality audio capture faces the challenge of complex noise environments. For example, during a musical performance, the performers' sound pickup equipment must not only capture the performers' delicate timbre, but also filter out interference from the audience's cheers, the sounds of musical instruments, and the stage's feedback speakers. Since the performers move on the stage, the sound environment for collecting sound changes dynamically with their position. Traditional fixed-position pickup and unified filtering strategies are difficult to adapt to this dynamic noise scenario, resulting in residual noise in the collected audio or distortion of the effective signal.

[0003] Existing technologies can usually only identify a single type of noise and are unable to generate a matching filtering strategy for the complex noise sources generated by position movement. The filter bandwidth is either too wide, resulting in loss of effective signals, or too narrow, resulting in incomplete noise filtering.

[0004] Therefore, how to determine the noise sources at different locations in real time, select the appropriate filtering algorithm, and improve the audio filtering effect has become an urgent problem that needs to be solved. Summary of the Invention

[0005] The present invention provides an adaptive filtering algorithm and device for a sound-intelligence membrane multi-noise source environment, which can determine the noise sources at different locations in real time, select the appropriate filtering algorithm, and improve the audio filtering effect.

[0006] In a first aspect, the present invention provides an adaptive filtering algorithm for a sound-intelligence membrane in a multi-noise source environment, comprising:

[0007] Analyze the physical structure of the performance scene within the sound-intelligence membrane to determine the multiple possible pickup points;

[0008] Determine the sound pickup environment corresponding to each sound pickup point in the sound intelligence membrane, and determine the corresponding noise source set based on the sound pickup environment;

[0009] The corresponding noise source set is acquired in real time at the point of the sound pickup device, and the corresponding filtering target is determined based on each noise source set to obtain the adaptive target after deduplication;

[0010] The corresponding filtering algorithm is determined based on the adaptive target, and the audio of the sound pickup device is filtered.

[0011] Optionally, in a possible implementation of the first aspect, the determining of multiple possible sound pickup points by analyzing the physical structure of the performance scene within the sound intelligence membrane includes:

[0012] Obtain a top view of the physical structure of the performance scene within the sound intelligence membrane, and identify the target object in the top view to obtain the target object's shape;

[0013] The target object shape is analyzed and processed to obtain a plurality of moving line trajectories, and the corresponding sound pickup points are determined based on the moving line trajectories.

[0014] Optionally, in a possible implementation of the first aspect, obtaining a top view of the physical structure of the performance scene within the sound intelligence membrane and identifying a target object in the top view to obtain a shape of the target object includes:

[0015] Obtain the coordinates of the physical structure of the performance scene within the sound intelligence film and identify the target object to obtain its coordinate points;

[0016] The coordinate points of the target object are regularized and identified to obtain multiple regular regions. All the regular regions are binarized and combined to obtain the binarized shape of the target object.

[0017] Optionally, in a possible implementation of the first aspect, the regularized recognition of the coordinate points of the target object to obtain a plurality of regular regions, and the binarization processing and combination of all the regular regions to obtain the binarized target object shape includes:

[0018] Obtain coordinate points having the same Y-axis coordinate and directly or indirectly adjacent X-axis coordinates to obtain a coordinate group, and count the number of coordinate points in the corresponding coordinate group to obtain a first coordinate number;

[0019] Count the Y-axis coordinates and horizontal coordinate intervals corresponding to each coordinate group;

[0020] If it is determined that the horizontal coordinate interval difference between adjacent Y-axis coordinates is less than or equal to a preset value, the corresponding coordinate group is divided into a regular area.

[0021] Optionally, in a possible implementation of the first aspect, if it is determined that the horizontal coordinate interval difference between adjacent Y-axis coordinates is less than or equal to a preset value, dividing the corresponding coordinate group into a regular area includes:

[0022] Extracting the first maximum value and the first minimum value of the adjacent first horizontal axis interval, and the second maximum value and the second minimum value of the adjacent second horizontal axis interval;

[0023] Calculate the difference between the first maximum value and the second maximum value to obtain the maximum difference, and calculate the difference between the first minimum value and the second minimum value to obtain the minimum difference;

[0024] The maximum difference and the minimum difference are converted to absolute values ​​and then added together to obtain the horizontal axis interval difference.

[0025] Optionally, in a possible implementation of the first aspect, generating and analyzing the target object shape to obtain a plurality of moving line trajectories, and determining corresponding sound pickup points based on the moving line trajectories includes:

[0026] Calculate the average length of each regular area in all X-axis directions;

[0027] The number of possible moving lines corresponding to the regular area is obtained by dividing the average length by the preset length and rounding it off.

[0028] Based on the number of moving lines, the regular area is evenly divided in all X-axis directions to obtain the moving lines as the average dividing lines. All moving lines are counted to obtain the moving line trajectories to determine the corresponding pickup points.

[0029] Optionally, in a possible implementation of the first aspect, counting all moving lines to obtain moving line trajectories and determining corresponding sound pickup points includes:

[0030] Determine the starting and ending points of each moving line as the pickup points;

[0031] If the length of the corresponding moving line is greater than the length of the split point, the number of pickups is obtained by dividing the moving line length by the length of the split point and rounding up.

[0032] The length of the moving line is evenly divided based on the number of pickups, and the evenly divided points are used as pickup points.

[0033] Optionally, in a possible implementation of the first aspect, determining a sound pickup environment corresponding to each sound pickup point within the sound intelligence membrane, and determining a corresponding noise source set based on the sound pickup environment, includes:

[0034] With the pickup point as the center point and the noise distance as the radius, the pickup environment corresponding to each center point is constructed in the upper view of the physical structure;

[0035] Extract noise features from each sound pickup environment in the top view of the physical structure to obtain a noise source set;

[0036] Each pickup point corresponds to a pickup environment and a set of noise sources.

[0037] Optionally, in a possible implementation of the first aspect, acquiring and determining corresponding noise source sets for the locations of the sound pickup devices in real time, and determining corresponding filtering targets based on each noise source set to remove duplicates and obtain an adaptive target include:

[0038] Obtain the position of the pickup device in real time and determine the closest pickup point;

[0039] Extract the noise source set of the closest pickup point and extract the real-time image to obtain the non-noise source;

[0040] The actual noise source set is obtained after deduplication of the noise source set based on the non-noise source, and the actual noise source set is compared with the preset filtering target to determine the corresponding filtering target, and each preset filtering target has a preset noise source combination.

[0041] A second aspect of the present invention provides an adaptive filtering device for a sound-intelligence membrane in a multi-noise source environment, comprising:

[0042] An analysis module is used to analyze the physical structure of the performance scene within the sound intelligence membrane to determine multiple possible pickup points;

[0043] A determination module, configured to determine the sound pickup environment corresponding to each sound pickup point within the sound intelligence membrane, and determine a corresponding noise source set based on the sound pickup environment;

[0044] An acquisition module is used to acquire the points of the sound pickup device in real time and determine the corresponding noise source set. Based on each noise source set, a corresponding filtering target is determined and duplicated to obtain an adaptive target;

[0045] The processing module is used to determine a corresponding filtering algorithm based on the adaptive target and perform filtering processing on the audio of the sound pickup device.

[0046] According to a third aspect of the present invention, a storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the method of the first aspect of the present invention and various possible designs of the first aspect.

[0047] The beneficial effects of the present invention are as follows:

[0048] 1. The present invention can determine the noise sources at different locations in real time, select appropriate filtering algorithms, and improve the audio filtering effect. First, the present invention can achieve scientific planning of pickup points and full scene coverage through stage entity structure analysis and dynamic trajectory modeling. Among them, the present invention first obtains the top view of the sound intelligence membrane performance scene, coordinates the stage target, extracts the regular area and calculates the horizontal coordinate interval difference, and divides it according to the preset length to reach the evenly divided dynamic line, thereby determining the pickup point and ensuring that all possible paths for people to move are covered, so as to improve the matching of the pickup point with the actual sound receiving position and avoid the instability of fixed position pickup.

[0049] 2. The present invention can achieve precise matching of filtering targets through sound pickup environment analysis and dynamic updating of noise source sets. The sound pickup environment is determined with the sound pickup point as the center and the noise distance as the radius. The noise sources within the environment are extracted and formed into a noise source set to automatically match the corresponding filtering targets. When the sound pickup device moves, the distance to the preset sound pickup point can be calculated in real time. The noise source set closest to the point is called and non-noise sources are eliminated. After deduplication, the actual filtering target is obtained, which facilitates improving the accuracy of identifying noise sources and improving the accuracy of filtering audio data. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of an adaptive filtering algorithm for a sound-intelligence membrane in a multi-noise source environment provided by the present invention;

[0051] Figure 2 A structural schematic diagram of a moving line provided by the present invention;

[0052] Figure 3 This is a schematic diagram of the structure of an adaptive filtering device for a multi-noise source environment provided by the present invention. DETAILED DESCRIPTION

[0053] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0054] like Figure 1 FIG. 1 is a flow chart of an adaptive filtering algorithm for a sound-intelligence membrane in a multi-noise source environment provided by the present invention. The adaptive filtering algorithm for a sound-intelligence membrane in a multi-noise source environment includes:

[0055] S1, analyze the physical structure of the performance scene within the sound intelligence membrane to determine the multiple possible pickup points.

[0056] It should be noted that since the microphones on the stage will be interfered with by various noises, such as the interference of the audience, sound and musical instruments, there will be multiple noise sources, and the interference frequency of each noise is also different, so multiple noise sources can be filtered. Since, when the filtering bandwidth is large, the sound of the corresponding performance will also be lost, and if the filtering bandwidth is too small, it will cause greater noise. Therefore, the pickup points can be identified first, so that the noise signal can be determined according to the corresponding pickup points, which is convenient for subsequent filtering.

[0057] It can be understood that the physical structure is the performance stage in the space of the sound-intelligence membrane, and the sound pickup points are the locations where the performers make sounds.

[0058] It is not difficult to understand that as people move around on the stage, the sound position will also change accordingly. Therefore, the pickup point where the sound is emitted can be determined so that the corresponding interfering noise can be determined later, which facilitates adaptive filtering processing.

[0059] In some embodiments, the specific implementation of step S1 (analyzing the physical structure of the performance scene within the sound intelligence membrane to determine the multiple possible sound pickup points) includes:

[0060] S11, obtaining a top view of the physical structure of the performance scene within the sound intelligence membrane, and identifying a target object in the top view to obtain the shape of the target object.

[0061] It can be understood that by flying the drone above the physical structure to take a bird's-eye view, a top view corresponding to the physical structure is obtained, and then the target object is identified in the top view to obtain the shape of the target object.

[0062] The top view is an image of the physical structure taken from above, the target object is a different stage part corresponding to the physical structure, and the target object shape is a shape corresponding to the target object, which may be a rectangle.

[0063] In some embodiments, a specific implementation of step S11 (obtaining a top view of the physical structure of the performance scene within the sound intelligence film and identifying a target object in the top view to obtain its shape) includes:

[0064] S111: Obtain the coordinates of the physical structure of the performance scene in the sound intelligence film, and identify the target object to obtain the coordinate points of the target object.

[0065] It is understandable that in order to determine the shape of the target object, the upper view may be subjected to coordinate processing so as to subsequently determine the contour coordinate points of the target object, thereby facilitating the subsequent determination of the corresponding target object shape.

[0066] The coordinate point is the coordinate point of the pixel corresponding to the target object.

[0067] It is not difficult to understand that the image center point of the top view can be selected as the coordinate origin, thereby realizing the coordinate processing of the top view.

[0068] S112 , performing regularized recognition on the coordinate points of the target object to obtain a plurality of regular regions, performing binarization processing on all the regular regions and combining them to obtain a binarized shape of the target object.

[0069] It is understandable that the coordinate points of the target object are usually discretely distributed. Directly using them for shape analysis will lead to inefficiency due to data redundancy. Regularized recognition forms geometrically regular areas by aggregating adjacent coordinate points, such as rectangles and strips, which can simplify shape features. Binarization processing strengthens the boundaries of the target object through maximum and minimum values, making it easier to determine the shape of the target object.

[0070] For example, the coordinate points of the singing area on the stage can form a rectangular area after regularization, and after binarization, the singing area can be clearly distinguished from other areas, providing a clear spatial range for subsequent movement trajectory analysis.

[0071] Through the above implementation, the generated binary target shape has a clear boundary and a uniform internal area, which is convenient for subsequent movement trajectory analysis.

[0072] In some embodiments, a specific implementation of step S112 (regularizing the coordinate points of the target object to obtain multiple regular regions, binarizing all the regular regions and combining them to obtain the binarized shape of the target object) includes:

[0073] S1121 , obtaining coordinate points having the same Y-axis coordinate and directly or indirectly adjacent X-axis coordinates to obtain a coordinate group, and counting the number of coordinate points in the corresponding coordinate group to obtain a first coordinate number.

[0074] It can be understood that in a two-dimensional coordinate system, points with the same Y-axis coordinate represent points on the same horizontal line. Points with direct or indirect X-axis proximity indicate that these points are continuously distributed horizontally and can form a horizontal line segment. Counting the number of first coordinates, that is, the number of points contained in the line segment, can reflect the length of the line segment.

[0075] For example, if the coordinate group contains 10 points, it means that the length of the horizontal line segment in the X-axis direction is the length of 10 coordinate points.

[0076] The coordinate group is a set of coordinates of points on the same horizontal line segment, and the first coordinate quantity is the total number of coordinate points in the coordinate group.

[0077] S1122: Count the Y-axis coordinates and the horizontal axis intervals corresponding to each coordinate group.

[0078] It can be understood that the Y-axis coordinate of each coordinate group determines its position in the vertical direction, and the horizontal coordinate interval is the coordinate interval corresponding to the horizontal coordinate, that is, the minimum to maximum value of the X-axis defines the horizontal coverage range of the horizontal line segment.

[0079] S1123: If it is determined that the horizontal coordinate interval difference between adjacent Y-axis coordinates is less than or equal to a preset value, the corresponding coordinate group is divided into a regular area.

[0080] It is understandable that if Figure 2 As shown in the figure, if the horizontal line segments of adjacent Y-axis coordinates have a small difference in the horizontal coordinate interval, it means that they are aligned in the horizontal direction and belong to the same regular area.

[0081] The preset value is a preset threshold used to determine whether a segment coordinate group belongs to the same regular area.

[0082] It is not difficult to understand that horizontal line segments are combined into two-dimensional regular areas, for example, multiple horizontal line segments aligned vertically are aggregated into a rectangular singing area, and the geometric error of the area division is less than or equal to the preset value, so as to provide clear area boundaries for subsequent binarization processing.

[0083] In some embodiments, a specific implementation of step S1123 (if it is determined that the horizontal coordinate interval difference between adjacent Y-axis coordinates is less than or equal to a preset value, then dividing the corresponding coordinate group into a regular area) includes:

[0084] S11231 , extracting the first maximum value and the first minimum value of the adjacent first horizontal axis interval, and the second maximum value and the second minimum value of the adjacent second horizontal axis interval.

[0085] It can be understood that the maximum and minimum values ​​of the first and second horizontal axis intervals are key parameters for describing the interval boundaries. Extracting these values ​​can clarify the left and right endpoint positions of the two intervals for subsequent calculation of the interval offset.

[0086] Among them, the first horizontal coordinate interval is one of the adjacent horizontal coordinate intervals, the second horizontal coordinate interval is another interval in the adjacent horizontal coordinate intervals, the first maximum value is the maximum value corresponding to the first horizontal coordinate interval, the first minimum value is the maximum value corresponding to the first horizontal coordinate interval, the second maximum value is the maximum value corresponding to the second horizontal coordinate interval, and the second minimum value is the minimum value corresponding to the second horizontal coordinate interval.

[0087] It is not difficult to understand that by extracting the maximum values ​​of adjacent intervals for subsequent comparison, the coordinate points corresponding to intervals with smaller differences are divided into the same regular area.

[0088] S11232: Calculate the difference between the first maximum value and the second maximum value to obtain a maximum difference, and calculate the difference between the first minimum value and the second minimum value to obtain a minimum difference.

[0089] It can be understood that the maximum difference is the difference between the maximum values ​​corresponding to adjacent horizontal coordinate intervals, that is, the difference between the first maximum value and the second maximum value, and the minimum difference is the difference between the minimum values ​​corresponding to adjacent horizontal coordinate intervals, that is, the difference between the first minimum value and the second minimum value.

[0090] It is not difficult to understand that the maximum difference reflects the offset of the right boundary of the two intervals, and the minimum difference reflects the offset of the left boundary. By decomposing it into the difference between the left and right boundaries, the difference value of the interval on both sides can be evaluated separately.

[0091] S11233, convert the maximum difference and the minimum difference into absolute values ​​and then add them together to obtain the horizontal axis interval difference.

[0092] It can be understood that the absolute value processing eliminates the influence of positive and negative signs, and the horizontal axis interval difference obtained after addition comprehensively reflects the total difference between the left and right sides.

[0093] For example, the maximum difference is 0.2 meters + the minimum difference is 0.3 meters = 0.5 meters. If the preset value is 0.6 meters, it is determined to be the same regular area.

[0094] S12, analyzing and processing the shape of the target object to obtain a plurality of moving line trajectories, and determining corresponding sound pickup points based on the moving line trajectories.

[0095] It is understandable that since the different shapes of the target objects affect the walking paths of the performers, the shape and position of the target objects can be analyzed to obtain multiple possible movement trajectories, so as to determine the corresponding pickup points of the performers based on the movement trajectories.

[0096] The moving line trajectory is the trajectory of the movement.

[0097] In some embodiments, the specific implementation of step S12 (generating and analyzing the target object shape to obtain multiple moving line trajectories, and determining corresponding sound pickup points based on the moving line trajectories) includes:

[0098] S121, counting the average length of each regular area in all X-axis directions.

[0099] It is understandable that in order to facilitate the subsequent determination of the corresponding moving line trajectory, the average length of the specification area on the X-axis can be calculated.

[0100] The average length is the horizontal spatial span of the regular area, so as to subsequently calculate the number of vertical movement lines that can be accommodated in the corresponding regular area.

[0101] S122: Based on the average length divided by the preset length and then rounded, the number of possible moving lines corresponding to the regular area is obtained.

[0102] It can be understood that the preset length is a pre-set division length, and the number of moving lines is the number of moving lines that can be accommodated in the regular area, that is, the integer value of the average length divided by the preset length.

[0103] It is not difficult to understand that the number of generated moving lines is proportional to the horizontal scale of the regular area, ensuring that the moving lines are evenly spaced. There will be no blind spots due to too few moving lines, nor will there be redundant pickup points due to too many moving lines, thus improving the rationality of the moving line distribution.

[0104] S123: Divide the regular area in all X-axis directions based on the number of moving lines to obtain moving lines as dividing lines, and count all moving lines to obtain moving line trajectories to determine corresponding sound pickup points.

[0105] It is understandable that if Figure 2 As shown in the figure, the regular area is divided equally according to the number of moving lines in the X-axis direction. The obtained average lines may be the paths of personnel activities. By counting the moving lines in all regular areas, a moving line trajectory network covering the entire performance scene can be formed. The endpoints and key nodes of the moving lines are the candidate locations of the pickup points.

[0106] The dividing line is a line segment that evenly divides the regular area.

[0107] In some embodiments, the specific implementation of step S123 (the step of counting all moving lines to obtain moving line trajectories and determining corresponding sound pickup points) includes:

[0108] S1231, determine the starting point and end point of each moving line as the sound pickup point.

[0109] It can be understood that the moving line is the line segment path closest to the person. Therefore, the starting point and end point of each moving line can be used as the pickup point. When the moving line is long, it can be divided accordingly to determine the corresponding pickup point.

[0110] S1232: If the length of the corresponding moving line is greater than the length of the split point, the number of picked-up sounds is obtained by dividing the moving line length by the length of the split point and rounding the result.

[0111] It is understandable that every point on the moving line cannot be a pickup point. Therefore, it is necessary to screen and determine the points on the moving line. When the length of the moving line is greater than the length of the dividing point, it can be said that the length of the moving line is long and exceeds the benchmark distance of the actual personnel movement point. Therefore, the moving line needs to be divided to determine the pickup points.

[0112] The length of the segmentation points is the pre-set length of the segmentation points, and the number of pickup points is the determined number of pickup points on the corresponding moving line.

[0113] It is not difficult to understand that this implementation method can flexibly determine the number of pickups according to the length of the moving line, avoiding the occurrence of sound collection blind spots due to the moving line being too long.

[0114] S1233: Divide the length of the moving line equally based on the number of picked-up sounds, and use the dividing points as the sound picking points.

[0115] It can be understood that by dividing the length of the moving line based on the number of pickups, the obtained dividing points can be evenly distributed on the moving line. These positions can collect sounds at different positions on the moving line more evenly, which is in line with the characteristics of sound propagation and adapts to the changes in sound when performers move on the moving line. The dividing points are used as pickup points, so that the sounds at various positions on the moving line can be better collected, reducing the collection errors caused by changes in sound position.

[0116] The dividing point is the point where the length of the moving line is divided equally.

[0117] S2: Determine the sound pickup environment corresponding to each sound pickup point in the sound intelligence membrane, and determine the corresponding noise source set based on the sound pickup environment.

[0118] It is understandable that different pickup points are located in different spatial positions, and the surrounding physical structures, such as equipment and audience distribution, will form a unique pickup environment. The types and intensities of noise sources in different environments vary. Determining the pickup environment corresponding to each pickup point and extracting the noise source set can enable subsequent filtering processing to accurately match the noise characteristics of each point, avoiding the poor effect caused by using a unified filtering strategy.

[0119] The sound pickup environment is the environment for collecting sounds, and the noise source set is the set corresponding to the noise sources.

[0120] Through the above-mentioned implementation, the present invention associates the sound pickup points with the noise source set so that the filtering algorithm can subsequently process the noise at each point in a targeted manner.

[0121] In some embodiments, a specific implementation of step S2 (determining the sound pickup environment corresponding to each sound pickup point within the sound intelligence membrane, and determining the corresponding noise source set based on the sound pickup environment) includes:

[0122] S21, with the pickup point as the center point and the noise distance as the radius, construct the pickup environment corresponding to each center point in the upper view of the physical structure.

[0123] It can be understood that the propagation of sound has a certain range. Therefore, in order to determine the noise that interferes with the pickup point, the pickup point can be used as the center point and the noise distance as the radius to determine the area corresponding to the interference of the corresponding pickup point, so as to determine the pickup environment corresponding to each center point in the upper view of the physical structure.

[0124] Among them, the noise distance is pre-set, that is, the distance that affects the sound pickup device to collect the performers' voices, for example, it can be 3 meters. Subsequently, the noise factors existing in the corresponding sound pickup environment can be extracted, such as the sounds emitted by the audience or musical instruments.

[0125] S22, extracting noise features from each sound pickup environment in the upper view of the physical structure to obtain a noise source set.

[0126] It can be understood that the features of noise generated in each sound pickup environment in the upper view of the physical structure are extracted, such as the people in the auditorium or the musical instruments corresponding to the equipment area, thereby forming a noise source set.

[0127] Through the above implementation, a noise source set corresponding to each sound pickup environment can be obtained, so that targeted filtering processing can be performed according to the characteristics of each noise source.

[0128] S23, each pickup point corresponds to a pickup environment and a set of noise sources.

[0129] It can be understood that the position of each pickup point is fixed and unique, and its corresponding pickup environment and noise source set are also unique. Establishing a one-to-one correspondence can ensure that during the pickup process, the noise source set information corresponding to the point can be quickly called to perform targeted filtering in real time, thereby shortening the filtering time and avoiding filtering errors caused by noise information confusion.

[0130] S3, acquiring and determining the corresponding noise source set for the point position of the sound pickup device in real time, determining the corresponding filtering target based on each noise source set and obtaining the adaptive target after deduplication.

[0131] It is understandable that the actual position of the pickup device may change dynamically due to performance requirements. Real-time acquisition of its position and matching with the closest preset pickup point can quickly associate the corresponding noise source set to determine the filtering target corresponding to the noise, making it easier to quickly filter the corresponding noise and obtain the required adaptive target.

[0132] Among them, the sound pickup device is a device that collects sound, the filtering target is the noise signal that needs to be filtered, and the adaptive target is the noise filtering signal determined by automatic matching.

[0133] It is not difficult to understand that when the position of the sound pickup device changes, the same noise source may exist in the noise source concentration corresponding to different sound pickup points. Therefore, the filter targets corresponding to the repeated noise sources can be deduplicated to obtain automatically matched filter targets for subsequent filtering of the collected audio data.

[0134] In some embodiments, a specific implementation of step S3 (acquiring and determining corresponding noise source sets for the locations of the sound pickup device in real time, and determining corresponding filtering targets for each noise source set and removing duplicates to obtain an adaptive target) includes:

[0135] S31, obtaining the position of the sound pickup device in real time and determining the closest sound pickup position.

[0136] It is understandable that the sound pickup device may deviate from the preset sound pickup point in actual use. After obtaining its point in real time, the closest sound pickup point is found by calculating the distance to each preset sound pickup point. The existing noise source set information at this point can be used to reduce the time for re-analyzing the noise source, so as to improve the filtering response speed.

[0137] S32: extract the noise source set of the closest sound pickup point, and extract the real-time image to obtain the non-noise source.

[0138] It is understandable that the noise source set closest to the sound pickup point is obtained based on historical data or preset situations, and may include noise sources that do not exist in the current environment. For example, the original sound pickup environment includes the auditorium, so the corresponding noise source set contains audience noise. However, when the sound is actually collected, it is extracted that there are no people in the auditorium in the corresponding area. In this case, the corresponding auditorium can be regarded as a non-noise source.

[0139] Among them, non-noise sources are targets that do not generate noise, such as auditoriums where no one is present.

[0140] S33, after deduplicating the noise source set based on the non-noise sources, an actual noise source set is obtained, and the actual noise source set is compared with the preset filtering targets to determine the corresponding filtering targets, where each preset filtering target has a preset noise source combination.

[0141] It can be understood that by using non-noise sources to deduplicate the noise source set, the noise sources that do not currently exist can be eliminated to obtain the actual noise source set, making it more consistent with the actual situation on site. The actual noise source set is compared with the preset filtering target. Since each preset filtering target corresponds to a specific noise source combination, the matching filtering target can be quickly found to ensure the correctness of the filtering direction.

[0142] It is not difficult to understand that different noise source sets have corresponding filtering targets, wherein the filtering targets are pre-set.

[0143] S4, determining a corresponding filtering algorithm based on the adaptive target, and performing filtering processing on the audio of the sound pickup device.

[0144] It is understandable that different adaptive targets correspond to different combinations of noise characteristics, and different filtering algorithms have different filtering effects on specific noises. Selecting a matching filtering algorithm based on the adaptive target can maximize the advantages of the algorithm, filter out the corresponding noise in a targeted manner, and reduce the loss of effective sound.

[0145] like Figure 3 FIG. 1 is a schematic diagram of the structure of an adaptive filtering device for a sound-intelligence membrane in a multi-noise source environment provided by the present invention. The adaptive filtering device for a sound-intelligence membrane in a multi-noise source environment comprises:

[0146] The analysis module is used to analyze the physical structure of the performance scene within the sound intelligence membrane to determine multiple possible pickup points.

[0147] The determination module is used to determine the sound pickup environment corresponding to each sound pickup point in the sound intelligence membrane, and determine the corresponding noise source set based on the sound pickup environment.

[0148] The acquisition module is used to acquire the points of the sound pickup device in real time and determine the corresponding noise source set. Based on each noise source set, the corresponding filtering target is determined and duplicated to obtain the adaptive target.

[0149] The processing module is used to determine a corresponding filtering algorithm based on the adaptive target and perform filtering processing on the audio of the sound pickup device.

[0150] The present invention also provides a storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the methods provided in the various embodiments described above.

[0151] The storage medium may be a computer storage medium or a communication medium. A communication medium includes any medium that facilitates the transfer of a computer program from one location to another. A computer storage medium may be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium may also be an integral part of the processor. The processor and the storage medium may be located in an application-specific integrated circuit (ASIC). In addition, the ASIC may be located in a user device. Of course, the processor and the storage medium may also exist as discrete components in a communication device. The storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0152] The present invention also provides a program product, which includes execution instructions stored in a storage medium. At least one processor of a device can read the execution instructions from the storage medium, and at least one processor executes the execution instructions so that the device implements the methods provided in the various embodiments described above.

[0153] In the above-mentioned terminal or server embodiments, it should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), etc. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly executed by a hardware processor or by a combination of hardware and software modules within the processor.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. The adaptive filtering algorithm of the Sound Intelligence Membrane for multiple noise source environments is characterized by: include: An analysis of the physical structure of the performance scene within the sound-intelligence membrane identified multiple possible pickup points, including: Obtain a top view of the physical structure of the performance scene within the sound intelligence membrane, and identify the target object in the top view to obtain the target object's shape; Analyzing and processing the shape of the target object to obtain multiple moving line trajectories, and determining corresponding sound pickup points based on the moving line trajectories; Determine the sound pickup environment corresponding to each sound pickup point in the sound intelligence membrane, and determine the corresponding noise source set based on the sound pickup environment; The corresponding noise source set is acquired in real time at the point of the sound pickup device, and the corresponding filtering target is determined based on each noise source set to obtain the adaptive target after deduplication; Determine the corresponding filtering algorithm based on the adaptive target and filter the audio of the pickup device; The generating and analyzing the target object shape to obtain a plurality of moving line trajectories, and determining corresponding sound pickup points based on the moving line trajectories, includes: Calculate the average length of each regular area in all X-axis directions; The number of possible moving lines corresponding to the regular area is obtained by dividing the average length by the preset length and rounding it off. Based on the number of moving lines, the regular area is evenly divided in all X-axis directions to obtain the moving lines as the average dividing lines. All moving lines are counted to obtain the moving line trajectories to determine the corresponding pickup points; The method of counting all moving lines to obtain moving line trajectories and determining corresponding pickup points includes: Determine the starting and ending points of each moving line as the pickup points; If the length of the corresponding moving line is greater than the length of the split point, the number of pickups is obtained by dividing the moving line length by the length of the split point and rounding up. Divide the length of the moving line equally based on the number of picked-up sounds, and use the equalization points as the pickup points; The determining of the sound pickup environment corresponding to each sound pickup point in the sound intelligence membrane, and determining the corresponding noise source set based on the sound pickup environment, includes: With the pickup point as the center point and the noise distance as the radius, the pickup environment corresponding to each center point is constructed in the upper view of the physical structure; Extract noise features from each sound pickup environment in the top view of the physical structure to obtain a noise source set; Each pickup point corresponds to a pickup environment and a set of noise sources; The method of acquiring and determining the corresponding noise source set from the point of the sound pickup device in real time, and determining the corresponding filtering target based on each noise source set and removing duplicates to obtain the adaptive target includes: Obtain the position of the pickup device in real time and determine the closest pickup point; Extract the noise source set of the closest pickup point and extract the real-time image to obtain the non-noise source; The actual noise source set is obtained after deduplication of the noise source set based on the non-noise source, and the actual noise source set is compared with the preset filtering target to determine the corresponding filtering target, and each preset filtering target has a preset noise source combination.

2. The adaptive filtering algorithm for multiple noise source environments of the sound intelligence membrane according to claim 1 is characterized in that: The step of obtaining a top view of the physical structure of the performance scene within the sound intelligence film and identifying a target object in the top view to obtain the shape of the target object includes: Obtain the coordinates of the physical structure of the performance scene within the sound intelligence film and identify the target object to obtain its coordinate points; The coordinate points of the target object are regularized and identified to obtain multiple regular regions. All the regular regions are binarized and combined to obtain the binarized shape of the target object.

3. The adaptive filtering algorithm for multiple noise source environments of the sound intelligence membrane according to claim 2 is characterized in that: The coordinate points of the target object are regularized to obtain a plurality of regular regions, and all the regular regions are binarized and combined to obtain the binarized shape of the target object, including: Obtain coordinate points having the same Y-axis coordinate and directly or indirectly adjacent X-axis coordinates to obtain a coordinate group, and count the number of coordinate points in the corresponding coordinate group to obtain a first coordinate number; Count the Y-axis coordinates and horizontal coordinate intervals corresponding to each coordinate group; If it is determined that the horizontal coordinate interval difference between adjacent Y-axis coordinates is less than or equal to a preset value, the corresponding coordinate group is divided into a regular area.

4. The adaptive filtering algorithm for multiple noise source environments of the sound intelligence membrane according to claim 3 is characterized in that: If it is determined that the horizontal coordinate interval difference between adjacent Y-axis coordinates is less than or equal to a preset value, the corresponding coordinate group is divided into a regular area, including: Extracting the first maximum value and the first minimum value of the adjacent first horizontal axis interval, and the second maximum value and the second minimum value of the adjacent second horizontal axis interval; Calculate the difference between the first maximum value and the second maximum value to obtain the maximum difference, and calculate the difference between the first minimum value and the second minimum value to obtain the minimum difference; The maximum difference and the minimum difference are converted to absolute values ​​and then added together to obtain the horizontal axis interval difference.

5. The adaptive filtering device for a sound-sensitive membrane multi-noise source environment according to any one of the algorithms of claims 1-4, characterized in that: include: An analysis module is used to analyze the physical structure of the performance scene within the sound intelligence membrane to determine multiple possible pickup points; A determination module, configured to determine the sound pickup environment corresponding to each sound pickup point within the sound intelligence membrane, and determine a corresponding noise source set based on the sound pickup environment; An acquisition module is used to acquire the points of the sound pickup device in real time and determine the corresponding noise source set. Based on each noise source set, a corresponding filtering target is determined and duplicated to obtain an adaptive target; The processing module is used to determine a corresponding filtering algorithm based on the adaptive target and perform filtering processing on the audio of the sound pickup device.

Images