Direct sound detection method, system and computer readable storage medium
By screening and covariance matrix decomposition in microphone array signals, the problem of poor sound source positioning effect in reverberation and noise environments is solved, and frequency points with high signal-to-noise ratio are obtained, which improves the accuracy of sound source positioning.
Patent Information
- Application Number
- CN202110767190.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-07-07
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2041-07-07
AI Technical Summary
In reverberation and noise environments, it is difficult for the prior art to effectively obtain direct sound signals with relatively high signal-to-noise, resulting in poor sound source positioning effect.
By receiving the array signal collected by the microphone array, obtaining the array frequency domain signal, filtering out the target frequency point with a signal-to-noise ratio higher than the threshold value, using the eigenvalue decomposition of the covariance matrix to filter out the direct signal energy greater than the reverberation signal energy Target frequency point to improve the accuracy of sound source positioning.
It achieves a frequency point with high signal-to-noise in reverberation and noise environments and is less affected by reverberation, which significantly improves the accuracy of sound source positioning.
Smart Images

Figure CN113687304B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of speech signal processing, and in particular to a direct sound detection method, system and computer-readable storage medium. Background Art
[0002] In daily life, acoustic equipment has been widely used in people's lives. When facing complex actual scenarios, the performance of a single microphone is often unsatisfactory. In order to achieve higher quality voice communication, we will also use microphone arrays to process voice signals. Sound source localization is a very important topic in microphone arrays, but the effect of sound source localization is often affected by reverberation and noise. Especially in a reverberant environment, the error of sound source localization will be very large, and commonly used sound source localization methods often cannot achieve the desired effect. When performing sound source localization in a reverberant and noisy environment, if a direct sound signal that is less contaminated by noise and reverberation can be obtained, the effect of sound source localization will be significantly improved.
[0003] The current method is to directly select the maximum power point on the power spectrum as the candidate point for estimating the azimuth. However, this method does not select the frequency point in combination with the noise spectrum, and does not consider the influence of noise and reverberation interference in actual scenes. Therefore, a new direct sound detection method is urgently needed to solve the above problems. Summary of the invention
[0004] The main technical problem solved by the present application is to provide a direct sound detection method, system and computer-readable storage medium, which can obtain frequency points with high signal-to-noise ratio and less impact of reverberation.
[0005] In order to solve the above technical problems, a technical solution adopted in the present application is: to provide a direct sound detection method, including: receiving an array signal acquired by a microphone array, and obtaining an array frequency domain signal based on the array signal; screening out a first target frequency point whose signal-to-noise ratio is higher than a first threshold value from the array frequency domain signal, and obtaining a first target signal at the location of the first target frequency point from the array frequency domain signal; obtaining a corresponding covariance matrix based on the first target signal, and performing eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix; screening out a second target frequency point whose direct signal energy is greater than the reverberation signal energy from the first target frequency point based on all the eigenvalues and the maximum eigenvalue, so as to use the second target frequency point to locate the sound source.
[0006] Among them, the step of screening out a second target frequency point in which the direct signal energy is greater than the reverberation signal energy from the first target frequency point according to all the eigenvalues and the maximum eigenvalue, and using the second target frequency point to localize the sound source includes: obtaining the ratio of the maximum eigenvalue to the sum of all the eigenvalues, and taking the first target frequency point corresponding to the ratio greater than the second threshold value as the second target frequency point; wherein the range of the ratio is 0-1.
[0007] Before the step of obtaining the ratio of the maximum eigenvalue to the sum of all the eigenvalues and taking the first target frequency point corresponding to the ratio greater than the second threshold value as the second target frequency point, the method includes: obtaining an index of a language function and a value of the language function according to the reverberation time; wherein the index is inversely proportional to the reverberation time; in response to the value of the language function being less than a first threshold value, taking the first threshold value as the second threshold value; in response to the value of the language function being greater than a second threshold value, taking the second threshold value as the second threshold value; in response to the value of the language function being greater than or equal to the first threshold value and less than or equal to the second threshold value, taking the value of the language function as the second threshold value.
[0008] Among them, the step of obtaining the corresponding covariance matrix according to the first target signal includes: obtaining the corresponding conjugate transposed matrix according to the first target signal; obtaining the expected value of the product of the first target signal and the conjugate transposed matrix, and using the expected value as the corresponding covariance matrix.
[0009] Among them, the step of performing eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix includes: decomposing the covariance matrix into an eigenvector matrix of the covariance matrix, a diagonal matrix composed of eigenvalues of the covariance matrix arranged in order from large to small, and a conjugate transposed matrix of the eigenvector matrix; obtaining all eigenvalues and the maximum eigenvalue of the covariance matrix according to the diagonal matrix, wherein the eigenvalues are the main diagonal elements of the diagonal matrix.
[0010] Among them, the steps of receiving the array signal acquired by the microphone array and obtaining the array frequency domain signal according to the array signal include: receiving the array signal acquired by the microphone array; performing frame processing and windowing operation on the array signal in sequence to obtain multiple time domain signals; converting the multiple time domain signals into multiple frequency domain signals by using fast Fourier transform; obtaining the sum of the product of the steering vector matrix and the sound source signal and the noise signal according to the frequency domain signal, and using the sum as the array frequency domain signal.
[0011] Among them, the step of filtering out a first target frequency point whose signal-to-noise ratio is higher than a first threshold value from the array frequency domain signal, and obtaining a first target signal at the position where the first target frequency point is located from the array frequency domain signal, includes: acquiring the frequency domain signal of one of the microphones in the array frequency domain signal, and obtaining the power of each frequency point in the frequency domain signal of the microphone; filtering out a third target frequency point whose power is greater than the first threshold value from the frequency domain signal of the microphone; filtering out a first target frequency point having the same position as the third target frequency point from the array frequency domain signal, and obtaining a first target signal at the position where the first target frequency point is located from the array frequency domain signal.
[0012] Among them, before the step of filtering out the third target frequency point whose power is greater than the first threshold value from the frequency domain signal of the microphone, it includes: obtaining the minimum power value of the frequency point within a preset time threshold according to the power, and using the minimum power value as the noise power; obtaining the product of the first multiple and the noise power, and using the product as the first threshold value.
[0013] In order to solve the above technical problems, another technical solution adopted in the present application is: to provide a direct sound detection system, including a memory and a processor coupled to each other, wherein the memory stores program instructions, and the processor is used to execute the program instructions to implement the direct sound detection method mentioned in any of the above embodiments.
[0014] In order to solve the above technical problems, another technical solution adopted in the present application is: providing a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is used to implement the direct sound detection method mentioned in any of the above embodiments.
[0015] Different from the prior art, the beneficial effect of the present application is as follows: in the present application, an array signal acquired by a microphone array is received, and an array frequency domain signal is obtained based on the array signal, a first target frequency point with a higher signal-to-noise ratio is screened out from the array frequency domain signal, and then the covariance matrix is decomposed by eigenvalue to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix, and a second target frequency point that is less affected by reverberation and dominated by direct sound is screened out from the first target frequency point based on all eigenvalues and the maximum eigenvalue, and finally a frequency point with a higher signal-to-noise ratio and less affected by reverberation can be obtained, thereby improving the accuracy of subsequent sound source positioning. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0017] in:
[0018] Figure 1 It is a flow chart of an implementation method of a direct sound detection method of the present application;
[0019] Figure 2 yes Figure 1 A schematic flow chart of an implementation method of step S1 in FIG.
[0020] Figure 3 yes Figure 1 A schematic flow chart of an implementation method of step S2 in FIG.
[0021] Figure 4 yes Figure 3 A schematic diagram of a flow chart of an implementation method before step S21;
[0022] Figure 5 yes Figure 1 A schematic flow chart of an implementation method of step 1 corresponding to step S3;
[0023] Figure 6 yes Figure 1 A schematic flow chart of an implementation method of step 1 corresponding to step S3;
[0024] Figure 7 yes Figure 1 A schematic diagram of a flow chart of an implementation method before the step corresponding to step S4;
[0025] Figure 8 It is a schematic diagram of the framework of an implementation method of a direct sound detection system of the present application;
[0026] Fig. 9 It is a structural schematic diagram of an implementation scheme of a direct sound detection system of the present application;
[0027] Fig.10 It is a schematic diagram of a framework of an implementation scheme of a computer-readable storage medium of the present application. DETAILED DESCRIPTION
[0028] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0029] See also Figure 1 , Figure 1 1 is a flow chart of an implementation method of a direct sound detection method of the present application. Specifically, the direct sound detection method includes:
[0030] S1: Receive the array signal acquired by the microphone array, and obtain the array frequency domain signal according to the array signal.
[0031] Specifically, in this embodiment, please refer to Figure 2 , Figure 2 yes Figure 1 Schematic diagram of a flow chart of an implementation method of step S1 in FIG. Specifically, step S1 includes:
[0032] S10: Receive array signals acquired by the microphone array.
[0033] S11: performing frame processing and windowing operations on the array signal in sequence to obtain multiple time domain signals.
[0034] S12: Convert the multiple time domain signals into multiple frequency domain signals using fast Fourier transform.
[0035] S13: Obtain the sum of the product of the steering vector matrix and the sound source signal and the noise signal according to the frequency domain signal, and use the sum as the array frequency domain signal.
[0036] Specifically, in this embodiment, a steering vector matrix, a sound source signal and a noise signal are obtained according to the frequency domain signal, and a frequency domain signal model is obtained according to the steering vector matrix, the sound source signal and the noise signal. Specifically, the frequency domain signal model is:
[0037] Array frequency domain signal: X(τ,f i )=A(f i ,θ)S(τ,f i )+N(τ,f i )
[0038] Steering vector matrix:
[0039] A(f i ,θ)=[a s1 (f i ,θ1),a s2 (f i,θ2),L,a sk+1 (f i ,θ k+1 )]∈C M*(k+1)
[0040] Sound source signal: S(τ,f i )=[s1(τ,f i ),s2(τ,f i ),L,s k+1 (τ,f i )] T ∈C (k+1)*1
[0041] Noise signal: N(τ,f i )=[n1(τ,f i ),n2(τ,f i ),L,n M (τ,f i )] T ∈C M*1
[0042] Array flow pattern:
[0043]
[0044] Among them, τ represents the delay τ required for the sound source to reach the first microphone, θ i Indicates that the i-th target is located at θ i Direction, f i Indicates the f i frequency points, d represents the array element spacing, T represents the transposition, there are k+1 targets and M array elements in total.
[0045] S2: Filter out a first target frequency point whose signal-to-noise ratio is higher than a first threshold value from the array frequency domain signal, and obtain a first target signal at a position where the first target frequency point is located from the array frequency domain signal.
[0046] Specifically, in this embodiment, please refer to Figure 3 , Figure 3 yes Figure 1 Schematic diagram of a flow chart of an implementation method of step S2 in FIG. Specifically, step S2 includes:
[0047] S20: Acquire a frequency domain signal of one of the microphones in the frequency domain signal of the array, and obtain the power of each frequency point in the frequency domain signal of the microphone.
[0048] Specifically, the frequency domain signal of one microphone in the microphone array is obtained, and the power P(t,f i ), where t and f i Indicates the frequency point f at time ti The power at the frequency point P(t,f i ) is a prior art and will not be described in detail in this application.
[0049] S21: Filter out a third target frequency point whose power is greater than a first threshold value from the frequency domain signal of the microphone.
[0050] Specifically, in this embodiment, please refer to Figure 4 , Figure 4 yes Figure 3 Schematic diagram of a flow chart of an implementation method before step S21. Specifically, before step S21, it includes:
[0051] S210: Obtain a minimum power value of a frequency point within a preset time threshold according to the power, and use the minimum power value as noise power.
[0052] Specifically, considering the non-stationarity of speech and the stationarity of noise, the minimum power value of each frequency point in the past 3 seconds is counted as the estimated noise power:
[0053] P noise (t,f i )=min(P(t0,f i )),t-3≤t0≤t
[0054] Of course, the above preset time threshold can also be set to other values according to user needs, and this application does not limit it here.
[0055] S211: Obtain a product of the first multiple and the noise power, and use the product as a first threshold value.
[0056] Specifically, in this embodiment, the first multiple is set to 3, using the noise power P noise (t,f i ) Calculate the first threshold: P th (f i )=3*P noise (t,f i ). Of course, in other embodiments, the first multiple may also be set to other values, which is not limited in this application.
[0057] Specifically, return to step S21, filter out the power P(t,f i ) is greater than the first threshold value P th (f i ) is the third target frequency point, denoted as f j .
[0058] S22: Filter out a first target frequency point at the same position as the third target frequency point from the array frequency domain signal, and obtain a first target signal at the position where the first target frequency point is located from the array frequency domain signal.
[0059] Specifically, from the array frequency domain signal X(τ,f i ) to select the third target frequency point f j The first target frequency point with the same position, and the frequency domain signal X(τ,f i ) to obtain the first target signal at the location of the first target frequency point, denoted as X(τ,f j ).
[0060] S3: Obtain a corresponding covariance matrix according to the first target signal, and perform eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and a maximum eigenvalue of the covariance matrix.
[0061] Specifically, in this embodiment, please refer to Figure 5 , Figure 5 yes Figure 1 Specifically, the step of obtaining the corresponding covariance matrix according to the first target signal in step S3 includes:
[0062] S30: Obtaining a conjugate transposed matrix corresponding to the first target signal according to the first target signal.
[0063] Specifically, according to the first target signal X(τ,f j ) to obtain the first target signal X(τ,f j ) corresponds to the conjugate transposed matrix X H (τ,f j ), where H represents the conjugate transpose.
[0064] S31: Obtain an expected value of the product of the first target signal and the conjugate transposed matrix, and use the expected value as the corresponding covariance matrix.
[0065] Specifically, according to the first target signal X(τ,f j ) and the conjugate transposed matrix X H (τ,f j ) Calculate the corresponding covariance matrix R(f j ):
[0066] R(f j )=E{X(τ,f j )X H (τ,f j )}
[0067] Where E{} represents the expected value.
[0068] Specifically, in this embodiment, please refer to Figure 6 , Figure 6 yes Figure 1 Specifically, in step S3, the step of performing eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix includes:
[0069] S40: decomposing the covariance matrix into an eigenvector matrix of the covariance matrix, a diagonal matrix formed by arranging the eigenvalues of the covariance matrix in descending order, and a conjugate transposed matrix of the eigenvector matrix.
[0070] Specifically, for the covariance matrix R(f j ) performs eigenvalue decomposition:
[0071] R(f j )=U(f j )Λ(f j ) H (f j )
[0072] Among them, U(f j ) represents the covariance matrix R(f j )’s eigenvector matrix; Λ(f j ) represents the diagonal matrix composed of the eigenvalues of the covariance matrix arranged in descending order, U H (f j ) represents the eigenvector matrix U(f j ) is the conjugate transposed matrix of .
[0073] S41: Obtain all eigenvalues and the maximum eigenvalue of the covariance matrix according to the diagonal matrix.
[0074] Specifically, the covariance matrix R(f j ) is the diagonal matrix Λ(f j )'s main diagonal elements, from all eigenvalues Pick out the largest eigenvalue
[0075] S4: selecting a second target frequency point where the direct signal energy is greater than the reverberation signal energy from the first target frequency point according to all eigenvalues and the maximum eigenvalue, so as to use the second target frequency point to perform sound source localization.
[0076] Specifically, in this embodiment, step S4 includes: obtaining the maximum eigenvalue With all eigenvalues The ratio of the sum of j : will be greater than the second threshold value rth The first target frequency point corresponding to the ratio of is taken as the second target frequency point. Specifically, the range of the above ratio is 0-1.
[0077] Considering that when there is only a single sound source, the energy of the reverberation signal incident on the array from a non-target angle is smaller than the direct signal, that is, the signal at the second target frequency point is the direct signal, and the rest of the signals are signals generated by reverberation. Then the energy of the direct signal is greater than the energy of the reverberation signal at some frequencies. According to all eigenvalues and the maximum eigenvalue A second target frequency point that is less affected by reverberation and is dominated by direct sound is selected from the first target frequency point. The second target frequency point is the frequency point dominated by direct sound. Using this frequency point for sound source localization will have better estimation performance, so that a frequency point with a higher signal-to-noise ratio and less affected by reverberation can be obtained, thereby improving the accuracy of subsequent sound source localization.
[0078] Specifically, in this embodiment, please refer to Figure 7 , Figure 7 yes Figure 1 Specifically, before the step of obtaining the ratio of the maximum eigenvalue to the sum of all eigenvalues and taking the first target frequency point corresponding to the ratio greater than the second threshold value as the second target frequency point, it includes:
[0079] S50: Obtaining an index of a language function and a value of the language function according to the reverberation time.
[0080] Specifically, the exponent of the above language function is related to the reverberation time T 60 Inversely proportional to In this embodiment, T 60 is the known reverberation time. Of course, in other embodiments, the reverberation time T 60 It can also be calculated by other methods. In addition, in this embodiment, the value of the language function is
[0081] S51: Determine the magnitude relationship between the value of the language function and the first threshold and the second threshold.
[0082] Specifically, in this embodiment, the first threshold and the second threshold are set to 0.1 and 0.8 respectively. Of course, the present application does not limit the values of the first threshold and the second threshold. The relationship between the first threshold and the second threshold is shown in the following formula:
[0083]
[0084] S52: When the value of the language function is less than the first threshold, the first threshold is used as the second threshold value.
[0085] Specifically, in this embodiment, when the value of the language function When it is less than 0.1, the second threshold value r th =0.1.
[0086] S53: When the value of the language function is greater than the second threshold, the second threshold is used as the second threshold value.
[0087] Specifically, in this embodiment, when the value of the language function When it is greater than 0.8, the second threshold value r th =0.8.
[0088] S54: When the value of the language function is greater than or equal to the first threshold and less than or equal to the second threshold, the value of the language function is used as the second threshold value.
[0089] Specifically, in this embodiment, when the value of the language function When the second threshold is greater than or equal to 0.1 and less than or equal to 0.8,
[0090] According to the reverberation time T 60 Calculate the second threshold value r th , when the reverberation is heavier, the second threshold value r th The higher the reverberation, the lower the second threshold value r th The lower the value, the lower the threshold value. th , thereby screening out the frequency points that are less affected by reverberation, thereby improving the accuracy of subsequent sound source positioning.
[0091] See also Figure 8 , Figure 8 Schematic diagram of a framework of an implementation method of a direct sound detection system of the present application. The direct sound detection system specifically includes:
[0092] The acquisition module 10 is used to receive an array signal acquired by the microphone array and acquire an array frequency domain signal according to the array signal.
[0093] The first screening module 12 is coupled to the acquisition module 10 and is used to screen out a first target frequency point whose signal-to-noise ratio is higher than a first threshold value from the array frequency domain signal.
[0094] The processing module 14 is coupled to the first screening module 12 and is used to obtain the first target signal at the location of the first target frequency point from the array frequency domain signal. Of course, the processing module 14 is also used to obtain the corresponding covariance matrix according to the first target signal, and perform eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix.
[0095] The second screening module 16 is coupled to the processing module 14 and is used to screen out a second target frequency point where the direct signal energy is greater than the reverberation signal energy from the first target frequency point according to all eigenvalues and the maximum eigenvalue, so as to use the second target frequency point for sound source localization.
[0096] See also Fig. 9 , Fig. 9 1 is a schematic diagram of a structure of an embodiment of a direct sound detection system of the present application. The direct sound detection system includes a memory 20 and a processor 22 coupled to each other. Specifically, in this embodiment, the memory 20 stores program instructions, and the processor 22 is used to execute the program instructions to implement the direct sound detection method mentioned in any of the above embodiments.
[0097] Specifically, the processor 22 may also be referred to as a CPU (Central Processing Unit). The processor 22 may be an integrated circuit chip having the ability to process signals. The processor 22 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. In addition, the processor 22 may be implemented by a plurality of integrated circuit chips.
[0098] See also Fig.10 , Fig.10It is a framework diagram of an implementation method of a computer-readable storage medium of the present application. The computer-readable storage medium 30 stores a computer program 300, which can be read by a computer, and the computer program 300 can be executed by a processor to implement the direct sound detection method mentioned in any of the above embodiments. Among them, the computer program 300 can be stored in the above-mentioned computer-readable storage medium 30 in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each implementation method of the present application. The computer-readable storage medium 30 with a storage function can be a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, RandomAccess Memory), a disk or an optical disk, etc., which can store program codes, or a terminal device such as a computer, a server, a mobile phone, and a tablet.
[0099] In summary, different from the prior art, the present application receives an array signal collected by a microphone array, obtains an array frequency domain signal based on the array signal, screens out a first target frequency point with a higher signal-to-noise ratio from the array frequency domain signal, and then uses the eigenvalue decomposition of the covariance matrix to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix. According to all the eigenvalues and the maximum eigenvalue, the second target frequency point that is less affected by reverberation and dominated by direct sound is screened out from the first target frequency point. Finally, a frequency point with a higher signal-to-noise ratio and less affected by reverberation can be obtained, thereby improving the accuracy of subsequent sound source positioning.
[0100] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.
Claims
1. A direct sound detection method, characterized in that: include: Receive an array signal acquired by the microphone array, and obtain an array frequency domain signal according to the array signal; Acquire a frequency domain signal of one of the microphones in the array frequency domain signal, and obtain the power of each frequency point in the frequency domain signal of the microphone; Obtaining a minimum power value of the frequency point within a preset time threshold according to the power, and using the minimum power value as the noise power; Obtaining a product of a first multiple and the noise power, and using the product as a first threshold value; Screening out a first target frequency point whose signal-to-noise ratio is higher than the first threshold value from the array frequency domain signal, and obtaining a first target signal at a position where the first target frequency point is located from the array frequency domain signal; Obtaining a corresponding covariance matrix according to the first target signal, and performing eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and a maximum eigenvalue of the covariance matrix; Obtaining a ratio of the maximum eigenvalue to the sum of all the eigenvalues, and taking a first target frequency point corresponding to the ratio greater than a second threshold value as a second target frequency point; wherein the range of the ratio is 0-1; The step of determining the second threshold value includes: Obtaining an index of a language function and a value of the language function according to the reverberation time; wherein the index is inversely proportional to the reverberation time; In response to the value of the language function being less than a first threshold, using the first threshold as the second threshold value; In response to the value of the language function being greater than a second threshold, using the second threshold as the second threshold value; In response to the value of the language function being greater than or equal to the first threshold and less than or equal to the second threshold, the value of the language function is used as the second threshold value.
2. The direct sound detection method according to claim 1, characterized in that: The step of obtaining a corresponding covariance matrix according to the first target signal comprises: Obtaining a corresponding conjugate transposed matrix according to the first target signal; An expected value of the product of the first target signal and the conjugate transposed matrix is obtained, and the expected value is used as the corresponding covariance matrix.
3. The direct sound detection method according to claim 2, characterized in that: The step of performing eigenvalue decomposition on the covariance matrix to obtain all eigenvalues and the maximum eigenvalue of the covariance matrix comprises: Decomposing the covariance matrix into an eigenvector matrix of the covariance matrix, a diagonal matrix formed by arranging the eigenvalues of the covariance matrix in descending order, and a conjugate transposed matrix of the eigenvector matrix; All eigenvalues and a maximum eigenvalue of the covariance matrix are obtained according to the diagonal matrix, wherein the eigenvalues are the main diagonal elements of the diagonal matrix.
4. The direct sound detection method according to claim 1, characterized in that: The step of receiving an array signal acquired by the microphone array and obtaining an array frequency domain signal according to the array signal comprises: Receive array signals acquired by the microphone array; Performing frame processing and windowing operations on the array signal in sequence to obtain multiple time domain signals; Converting the multiple time domain signals into multiple frequency domain signals using fast Fourier transform; The sum of the product of the steering vector matrix and the sound source signal and the noise signal is obtained according to the frequency domain signal, and the sum is used as the array frequency domain signal.
5. The direct sound detection method according to claim 1, characterized in that: The step of screening out a first target frequency point having a signal-to-noise ratio higher than a first threshold value from the array frequency domain signal, and obtaining a first target signal at a position where the first target frequency point is located from the array frequency domain signal, comprises: Acquire a frequency domain signal of one of the microphones in the array frequency domain signal, and obtain the power of each frequency point in the frequency domain signal of the microphone; Filter out the third target frequency point whose power is greater than the first threshold value from the frequency domain signal of the microphone; A first target frequency point having the same position as the third target frequency point is screened out from the array frequency domain signal, and a first target signal at a position where the first target frequency point is located is obtained from the array frequency domain signal.
6. A direct sound detection system, characterized in that: It comprises a memory and a processor coupled to each other, wherein the memory stores program instructions, and the processor is used to execute the program instructions to implement the direct sound detection method according to any one of claims 1 to 5.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is used to implement the direct sound detection method according to any one of claims 1 to 5.
Citation Information
Patent Citations
Sound source positioning system based on distributed microphone array
CN107102296A
Speech recognition method and device based on artificial intelligence
CN107316649A
