Sound source localization and speech enhancement methods and devices

Abstract

This invention discloses a sound source localization and speech enhancement method and apparatus. The speech enhancement method includes: converting dual-microphone signals to the frequency domain and converting the frequency domain signals to polar coordinates; calculating the slope range of the phase difference of the microphone signals based on the microphone spacing and sound speed, and dividing the phase difference slope according to accuracy requirements; converting each divided phase difference slope into phase compensation to the second microphone spectrum, and calculating the difference matrix and summation matrix; obtaining the best matching slope based on the difference matrix and summation matrix; expanding the best matching slope and combining it with the target sound source azimuth range to obtain the target slope interval; calculating the compensation spectrum of the second microphone based on the target slope interval, and performing target speech enhancement with the spectrum signal of the first microphone; calculating the gain factor by blocking the spectrum in the target direction and non-target direction; and further suppressing sound sources in non-target direction by comparing the relationship between the phase difference of the sound source and the phase difference in the target direction.

Description

Technical Field

[0001] This invention relates to the field of audio signal processing technology, and in particular to a method and apparatus for sound source localization and speech enhancement. Background Technology

[0002] In daily life, people can distinguish the direction and content of sounds of interest from a noisy environment. This is due to both the unique structure of the human ear and the phase difference that occurs when sound reaches the ear, as well as the brain's analysis of the signal. In the field of sound signal processing, similar functions can be achieved using multi-microphone arrays and corresponding algorithms, namely, locating and enhancing sound sources.

[0003] Sound source localization technology is applied in various fields, such as military, marine search and rescue, multimedia communication, and intelligent robotics. Based on the different transmitted and received signals, sound source localization can be divided into active sound source localization and passive sound source localization. Active sound source localization involves a signal generating device actively emitting a signal and determining the location of an object by the reflected signal, such as radar and sonar. Passive sound source localization only passively receives sound source signals and determines the direction of the sound source, such as sound source localization in a conference system. This invention mainly discusses passive sound source localization.

[0004] Passive sound source localization relies on the phase difference of the microphone array signal. The delay of the microphone array signal is calculated from the phase difference, and then the location of the sound source is calculated using trigonometric relationships, such as... Figure 1 As shown, A and B are the positions of two microphones, separated by a distance L. A sound source (assumed to be a far-field source) is transmitted from the far end to the two microphones in the form of a plane wave. Finding the position of the sound source can be considered as finding the magnitude of θ. θ is the time difference between the arrival of the signal at microphones A and B; once this time difference is found, the magnitude of θ can be determined using trigonometric relationships.

[0005] The general algorithm calculates the correlation between the two microphones, assuming the signals arriving at the two microphones are:

[0006] r1(t)=s(t)+n1(t.) (1)

[0007] r2(t)=s(tD)+n2(t) (2)

[0008] Where r1(t) and r2(t) represent the total signals received by the two microphones, including the signals s(t) and s(tD) emitted by the sound source, and the random noise n1(t) and n2(t) received by the two microphones. The cross-correlation and delay of the two microphone signals are:

[0009] R(τ)=E{r1(t)r2(t+τ)} (3)

[0010]

[0011] Equation (3) is the cross-correlation expression for the microphone signals. Once the microphone signals are determined, the magnitude of this value is related to the delay. When the delay equals the actual delay, equation (3) has its maximum value. After the delay is calculated, the angle θ can be obtained using θ = arcsin(cτ / L), where L is the distance between the two microphones and c is the speed of sound.

[0012] However, the following problems can be found from the above derivation:

[0013] (1) Since the time-domain signal is a series of discrete values, the precision of the value is the reciprocal of the sampling rate, which affects the precision of θ. When the microphone spacing L is constant, the higher the sampling rate, the higher the precision of θ. When L is 5cm and the sampling rate is 16000, assuming the speed of sound is 343m / s, the precision of the θ value is 25.4°. This is not a good precision, and the performance will be worse as the microphone distance L decreases. If it is necessary to improve the precision, upsampling is required, which wastes computing resources.

[0014] (2) The time-domain cross-correlation method used in this method has a complexity of O(n^2). 2 This results in a large computational burden when the sequence is very long.

[0015] Furthermore, once the sound source's location is calculated, a multi-microphone array can utilize beamforming algorithms to amplify sound from the target direction while suppressing sound from other directions. Beamforming algorithms leverage the phase information generated by the same sound source arriving at different microphones to amplify sound from a specific direction. For example... Figure 1 As shown, taking two microphones as an example, the arrival time difference can be determined based on the target direction θ and the distance L between the two microphones A and B.

[0016] τ=sinθ*L / c (5)

[0017] Where c is the speed of sound. Once the time difference is confirmed, a commonly used speech enhancement method is delay and sum, which involves aligning the microphone signals and then adding them together, such as... Figure 2 As shown. In the case of two microphones, this algorithm can theoretically suppress 3dB of ambient noise; the more microphones, the stronger the suppression of ambient noise, without jeopardizing the target signal. However, this speech enhancement method also has the following drawbacks:

[0018] (1) At typical sampling rates (8k, 16k, 32k, 48k), it is difficult to accurately align two microphone signals. For example, at an 8k sampling rate, the minimum time interval between sampling points is 1 / 8000 seconds. If the time difference calculated by equation (5) is not an integer multiple of this value, the two microphone signals cannot be perfectly aligned, which in turn affects the effectiveness of the algorithm.

[0019] (2) The method can theoretically suppress 3dB of ambient noise in the case of two microphones, but in fact the suppression strength is not high and cannot meet certain specific needs. Summary of the Invention

[0020] To overcome the shortcomings of the existing technology, one objective of the present invention is to provide a sound source localization and speech enhancement method and apparatus, so as to achieve the purpose of finding the location of the sound source through sound source localization.

[0021] Another objective of this invention is to provide a sound source localization and speech enhancement method and apparatus, which locates the sound source location through sound source localization, determines the target location range, enhances the sound at the target location, and suppresses the sound at non-target locations.

[0022] Another objective of this invention is to provide a sound source localization and speech enhancement method and apparatus, which can enhance sound from a fixed direction.

[0023] To achieve the above objectives, the present invention provides a sound source localization method, comprising the following steps:

[0024] Step S1: Convert the dual-microphone signal to the frequency domain and convert the frequency domain signal to polar coordinates.

[0025] Step S2: Calculate the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and divide the phase difference slope according to the accuracy requirements;

[0026] Step S3: Convert the slopes of each phase difference into the spectrum of the second microphone after phase compensation, and calculate the difference matrix and summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone.

[0027] Step S4: Calculate the best matching slope based on the difference matrix and the summation matrix;

[0028] Step S5: Based on the obtained best matching slope, the delay of the signal arriving at the microphone is deduced, and the signal orientation is calculated using trigonometric relationships, which is then used as the orientation output.

[0029] Optionally, in step S2, the maximum phase difference slope of the first microphone and the second microphone is calculated based on the microphone spacing and sound speed, the phase difference slope range is determined based on the maximum phase difference slope, and the phase difference slope is divided equally according to the accuracy requirements.

[0030] Optionally, step S3 further includes:

[0031] The phase differences represented by each slope are sequentially compensated to the second microphone, and the compensated spectrum of the second microphone is recovered from the compensated phase.

[0032] Subtracting the phase-compensated spectrum of the second microphone from the spectrum of the first microphone yields the MD. m (k) Add the phase compensation spectrum of the second microphone to the spectrum of the first microphone to obtain PD. m (k), and finally MD respectively m (k) and PD m (k) Summing the absolute values ​​of the frequency points yields the difference matrix and the summation matrix.

[0033] Optionally, in step S4, the difference matrix and the summation matrix are summed according to the number of frames to obtain two vectors, and then the two vectors are compared, and the phase difference slope corresponding to the smallest ratio is selected as the best matching slope.

[0034] To achieve the above objectives, the present invention also provides a method for enhancing speech from a sound source, comprising the following steps:

[0035] Step S1: Convert the dual-microphone signal to the frequency domain and convert the frequency domain signal to polar coordinates.

[0036] Step S2: Calculate the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and divide the phase difference slope according to the accuracy requirements;

[0037] Step S3: Convert the slopes of each phase difference into the spectrum of the second microphone after phase compensation, and calculate the difference matrix and summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone.

[0038] Step S4: Calculate the best matching slope based on the difference matrix and the summation matrix;

[0039] Step S5: Expand the best matching slope and combine it with the specified target sound source azimuth range to obtain the target slope range;

[0040] Step S6: Calculate the compensation spectrum of the second microphone based on the target slope range, and enhance the target speech by combining it with the spectrum signal of the first microphone using a delay accumulation algorithm.

[0041] Optionally, the method further includes:

[0042] Step S7: Calculate the blocking spectrum of the target signal and the non-target signal based on the best matching slope and the target slope interval, and use the blocking spectrum to construct the result of the first gain applied to the delay accumulation algorithm.

[0043] Optionally, step S7 further includes:

[0044] Step S700: Calculate the compensation spectrum of the two end-fire directions, and obtain the blocking spectrum of the two end-fire directions based on the compensation spectrum, and obtain the first blocking spectrum based on the blocking spectrum of the two end-fire directions.

[0045] Step S701, if the best matching slope is located in the target slope interval [b m1 b m2 Within [ ], the blocking spectrum with the best matching slope is calculated, and its absolute value is taken as the second blocking spectrum;

[0046] Step S702: If the best matching slope is outside the target slope interval, calculate the blocking spectrum of the best matching slope, update the first blocking spectrum based on the blocking spectrum of the best matching slope b and the first blocking spectrum, and at the same time, search for the best matching slope again in the target slope interval, and calculate the corresponding blocking spectrum amplitude as the second blocking spectrum.

[0047] Step S703: Based on the first blocking spectrum and the second blocking spectrum, construct the result of the first gain factor applied to the delay accumulation algorithm.

[0048] Optionally, the method further includes:

[0049] Step S8: The second gain factor is determined by comparing the spectral phase difference between the first microphone and the second microphone with the phase difference interval corresponding to the target slope interval, so as to suppress signals outside the target phase difference.

[0050] To achieve the above objectives, the present invention also provides a sound source localization device, comprising:

[0051] The time-frequency conversion and polar coordinate processing unit is used to convert the dual microphone signals to the frequency domain and convert the frequency domain signals into polar coordinate form;

[0052] The phase difference slope division unit is used to calculate the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and to divide the phase difference slope according to the accuracy requirements.

[0053] The phase compensation unit is used to convert the slopes of each phase difference into the spectrum of the second microphone with phase compensation, and to calculate the difference matrix and the summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone.

[0054] The optimal matching slope calculation unit is used to calculate the optimal matching slope based on the difference matrix and the summation matrix.

[0055] The azimuth estimation unit infers the delay of the signal arriving at the microphone based on the obtained best-matching slope, and uses trigonometric relationships to calculate the signal azimuth as the azimuth output.

[0056] To achieve the above objectives, the present invention also provides a sound source speech enhancement device, comprising:

[0057] The time-frequency conversion and polar coordinate processing unit is used to convert the dual microphone signals to the frequency domain and convert the frequency domain signals into polar coordinate form;

[0058] The phase difference slope division unit is used to calculate the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and to divide the phase difference slope according to the accuracy requirements.

[0059] The phase compensation unit is used to convert the slopes of each phase difference into the spectrum of the second microphone with phase compensation, and to calculate the difference matrix and the summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone.

[0060] The optimal matching slope calculation unit is used to calculate the optimal matching slope based on the difference matrix and the summation matrix.

[0061] The slope extension unit is used to extend the best matching slope and combine it with the specified target sound source azimuth range to obtain the target slope range.

[0062] The spectrum compensation unit is used to calculate the compensation spectrum of the second microphone based on the target slope range, and to enhance the target speech by combining it with the spectrum signal of the first microphone using a delay accumulation algorithm.

[0063] Compared with existing technologies, this invention provides a sound source localization and speech enhancement method and apparatus. It locates the sound source's position through sound source localization, then determines the target location range, and simultaneously enhances the sound from the target location while suppressing sounds from non-target locations. Verification has shown that this invention can enhance the target sound source while suppressing sounds from non-target directions, and it can also enhance speech in far-field and reverberant environments. Furthermore, this invention has the advantage of low computational complexity and can easily run on common CPUs.

[0064] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0065] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0066] Figure 1 This is a schematic diagram of the orientation estimation using existing dual-microphone technology;

[0067] Figure 2 This is a schematic diagram illustrating the enhancement of the delay and sum algorithm in existing technologies.

[0068] Figure 3 This is a flowchart illustrating the sound source localization method provided in the first exemplary embodiment of the present invention;

[0069] Figure 4 This is a schematic flowchart of a sound source speech enhancement method provided in a second exemplary embodiment of the present invention;

[0070] Figure 5 This is a detailed flowchart of the dual-microphone voice enhancement process in this embodiment;

[0071] Figure 6 This is a diagram illustrating the dual-microphone voice enhancement effect according to an embodiment of the present invention;

[0072] Figure 7 This is a system structure diagram of the sound source localization device provided in the third exemplary embodiment of the present invention;

[0073] Figure 8 This is a system structure diagram of the sound source speech enhancement device provided in the fourth exemplary embodiment of the present invention.

[0074] Figure 9 This is the structure of an electronic device provided in an exemplary embodiment of the present invention. Detailed Implementation

[0075] The following describes the embodiments of the present invention through specific examples and in conjunction with the accompanying drawings. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific examples, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0076] Example 1:

[0077] Figure 3This is a schematic flowchart of the sound source localization method provided in the first exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as... Figure 3 As shown, it includes the following steps:

[0078] Step S1: Convert the dual-microphone signal to the frequency domain and convert the frequency domain signal to polar coordinates.

[0079] Specifically, step S1 further includes

[0080] Step S101: Convert the two microphone signals into frequency domain signals.

[0081] Specifically, the time-domain signals d1 and d2 from the two microphones are framed and subjected to Discrete Fourier Transform (DFT) to obtain the frequency-domain signals D1 and D2.

[0082] Step S102: Obtain the polar coordinate representation of the spectrum of each microphone signal.

[0083] In this embodiment, each frequency point of the frequency domain signals D1 and D2 (corresponding to the first microphone 1 and the second microphone 2, respectively) is divided into real and imaginary parts, which can be converted into polar coordinates:

[0084] absD1=abs(D1) (6)

[0085] absD2=abs(D2) (7)

[0086] angleD1=angle(D1) (8)

[0087] angleD2=angle(D2) (9)

[0088] Where absD1 and absD2 represent the frequency domain amplitudes of the two microphones, and angleD1 and angleD2 represent the frequency domain angles of the two microphones.

[0089] If the two microphones are highly consistent, and random noise from the microphones is eliminated, then the following relationship should hold:

[0090] absD1(k)=absD2(k) (10)

[0091] angleD1(k)-angleD2(k)=b*k (11)

[0092] Where k represents the frequency index and b is a constant.

[0093] Equation (10) above indicates that, under the condition of good microphone consistency, the amplitude of each frequency point of the two microphones is the same; Equation (11) indicates that the phase difference of each frequency point of the two microphones is linear. If b is 0, it means that the phases of the two microphone signals are completely consistent, that is, the sound source arrives at the two microphones at the same time. The larger the absolute value of b, the larger the phase difference between the two microphones, that is, the greater the delay of the sound source arriving at the two microphones. And the value of b corresponds one-to-one with the delay. Therefore, finding the delay can be equivalent to finding the slope b of the phase difference between the two microphone signals.

[0094] Step S2: Calculate the maximum phase difference slope of the microphone signal based on the microphone spacing, and divide the maximum phase difference slope according to the accuracy requirements.

[0095] In this embodiment, step S2 further includes:

[0096] Step S200: Calculate the range of the phase difference slope between the two microphones based on the distance between them.

[0097] Given that the microphone distance is L, for sound waves with wavelength λ equal to L, the maximum phase difference is 2π. For sound waves of other wavelengths, the maximum phase difference satisfies:

[0098]

[0099] There is also a phase difference.

[0100]

[0101] in fs is the sampling rate, N is the FFT length, and c is the speed of sound. Substituting these values ​​into equation (12), and combining equations (12) and (13), we can obtain...

[0102] B=(fs*L*2π) / (N*c) (14)

[0103] In equation (14), B is the absolute value of the slope of the linear phase difference generated by the sound source in the direction of the two microphones when the microphone distance is fixed. The actual absolute value of the slope should be less than or equal to this value.

[0104] Step S201: Divide the phase difference slope equally according to the accuracy requirements.

[0105] In this embodiment, the obtained phase difference slope value ranges from -B to B. The slope value from -B to B is divided into 2M+1 values: M positive values, M negative values, and a 0 value. The evenly divided slope is represented as a vector {b0, b1, ..., b...} 2M}

[0106] Step S3: Convert the slopes of each phase difference into the spectrum of the second microphone after phase compensation, and calculate the difference matrix and summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone.

[0107] Specifically, the phase differences represented by each slope are sequentially compensated for by the second microphone 2, and the compensated spectrum of the second microphone 2 is reconstructed from the compensated phase:

[0108]

[0109]

[0110] in The spectrum of the second microphone 2 after phase compensation is represented, including the real and imaginary parts. The value of is related to m, representing the phase compensation of each equal division.

[0111] Then, subtract the phase compensation spectrum of the second microphone 2 from the spectrum of the first microphone 1. Get MD m (k) The spectrum of the first microphone 1 is added to the phase-compensated spectrum of the second microphone 2. Get PD m (k), and finally MD respectively m (k) and PD m (k) Summing the absolute values ​​of the frequency points yields the SMD. m SPD m SMD m With SPD m Given two vectors of length 2M+1, store each vector in a matrix MSMD. m With MSPD m The end of it is as follows:

[0112]

[0113]

[0114] SMD m =∑ k |MD m (k)| (19)

[0115] SPD m =∑ k |PD m (k)| (20)

[0116]

[0117]

[0118] Equations (19) and (20) represent summing the absolute values ​​of frequencies, where the summation range can be selective or full-band. This yields the SMD. m and SPD m It is a 2M+1 dimensional vector, representing the sum of the spectral amplitudes after differential and summation of the two microphones under different delay compensations. When the phase-compensated spectrum of the second microphone 2 matches the spectrum of the first microphone 1, SMD m and SPD m Theoretically, they should each have a minimum and a maximum value. Equations (21) and (22) represent the SMD m and SPD m Put them into the matrix MSMD respectively m and MSPD m MSMD m For difference matrices, MSPD m To calculate the summation matrix, the dimension of the matrix in this embodiment is (L+1)*(2M+1). This is done to take into account the phase information of (L+1) frames at the same time, thereby increasing the robustness of the positioning algorithm.

[0119] Step S4: Calculate the best matching slope based on the difference matrix and the summation matrix.

[0120] In this embodiment, the difference matrix MSMD m Summation matrix MSPD m Summing the values ​​based on the frame count yields two 2M+1 dimensional vectors. The ratio of these two vectors is then calculated, and the b-value corresponding to the smallest ratio is selected as the optimal matching slope, as detailed below:

[0121] SMSMD m =∑ l MSMD m (twenty three)

[0122] SMSPD m =∑ l MSPD m (twenty four)

[0123] RMP m =SMSMD m . / SMSPD m (25)

[0124] b = min m (RMP m (26)

[0125] Among them, RMP m It is a 2M+1 dimensional vector, which has a minimum value when the phases are matched. b is the slope that makes the phases of the two microphone signals most matched after phase compensation, and the microphone delay can be deduced from the value of b.

[0126] Step S5: Based on the obtained best matching slope, the delay of the signal arriving at the microphone is deduced, and the signal orientation is calculated using trigonometric relationships, which is then used as the orientation output.

[0127] In this embodiment, the formula for calculating the microphone delay is derived from the b-value as follows:

[0128] τ=bN / (fs*2π) (27)

[0129] Once the delay τ is calculated, the angle of the sound source can be calculated based on the microphone spacing and the speed of sound, which can then be used as the output for sound source localization.

[0130] Example 2:

[0131] As can be seen from the aforementioned embodiment one, the slope of equation (26) can be used to calculate the compensation phase of the microphone using equation (15), and then the compensation spectrum D′2(k) of the second microphone 2, including the real and imaginary parts, can be calculated using equation (16). Ideally, the compensation spectrum D′2(k) of the second microphone 2 is consistent with the spectrum D1(k) of the first microphone 1. However, due to microphone consistency, microphone background noise, and other reasons, only the spectrum of the sound source in the target direction is consistent. The delay and sum algorithm can be used to enhance the target speech:

[0132] D out (k)=(D1(k)+D′2(k)) / 2 (28)

[0133] For signals in the target direction, the summed value should be the same as the original value; for signals not in the target direction, the summed value should be less than twice the original value. (D) out (k) Perform IDFT transformation to obtain the final output d of the delay and sum algorithm. out (t).

[0134] Compared with time-domain delay and sum, this method is not limited by the sampling rate and can perfectly align the signals from two microphones, achieving speech enhancement to the theoretical limit of the algorithm. However, in far-field situations or when reverberation is severe, the phase relationship of the sound sources is relatively ambiguous, and the slope b obtained from equation (26) may not be accurate. Therefore, the improvement of the signal-to-noise ratio will be affected, and it may also cause loss to far-field speech. In order to protect far-field speech, this embodiment extends the value of the slope b and imposes certain constraints on the extended value.

[0135] Figure 4 This is a schematic flowchart of a sound source speech enhancement method provided in a second exemplary embodiment of the present invention. Figure 5This is a detailed flowchart of the dual-microphone voice enhancement process in this embodiment. This embodiment can be applied to electronic devices, such as... Figure 4 and Figure 5 As shown, it includes the following steps:

[0136] Step S-1: Convert the dual-microphone signal to the frequency domain and convert the frequency domain signal to polar coordinates.

[0137] Step S-2: Calculate the maximum phase difference slope of the microphone signal based on the microphone spacing, and divide the maximum phase difference slope according to the accuracy requirements.

[0138] Step S-3: Convert the slopes of each phase difference into the spectrum of the second microphone after phase compensation, and calculate the difference matrix and summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone.

[0139] Step S-4: Calculate the best matching slope based on the difference matrix and the summation matrix.

[0140] Since steps S-1 to S-4 are the same as in Embodiment 1, they will not be repeated in this embodiment.

[0141] Step S-5: Expand the best matching slope and combine it with the specified target sound source azimuth range to obtain the target slope range.

[0142] When two microphones are symmetrically distributed, the sound within a certain angle range directly in front is usually amplified. The sound source in the end-firing direction of the two microphones will not be the target sound source, so a strategy of suppressing the sound source in the end-firing direction can be adopted. Of course, the pickup angle range can also be freely selected according to the correspondence between the slope b and the angle. As can be seen from the above formula (14), the value range of the slope b is [-B, B], and the value of b corresponds one-to-one with the angle. When the target sound source angle range is determined, the value range of the slope b is also uniquely determined, assumed to be [b1, b2]. The above formula (26) determines the best matching slope b. In order to protect the far-field sound source, this embodiment extends the slope b to [b-δ, b+δ], and takes the intersection with [b1, b2] as the final target slope interval [b1, b2]. m1 b m2 ].

[0143] Step S-6: Calculate the compensation spectrum of the second microphone 2 based on the target slope range, and enhance the target speech by combining it with the spectrum signal of the first microphone 1 using a delay and sum algorithm.

[0144] Assuming the number of slopes within the interval is mm, the target slope interval [b] is... m1 b m2Substituting into the aforementioned formulas (15) and (16), we obtain the second microphone 2 compensation spectrum D′2(k, m) for the series target range. From the m compensation spectra, we select the real and imaginary parts with larger amplitudes, i.e.:

[0145] real(D′2(k))=max(|real(D′2(k,m))|) / |real(D′2(k,m))|*real(D′2(k,m))(29)

[0146] imag(D′2(k))=max(|imag(D′2(k,m))|) / |imag(D′2(k,m))|*imag(D′2(k,m))(30)

[0147] Once D′2(k) is obtained, the delay and sum output can be obtained by using the aforementioned formula (28), which can solve the speech loss problem in environments such as far field and reverberation.

[0148] Optionally, to suppress non-target sound sources, the speech enhancement method of this embodiment further includes:

[0149] Step S-7: Calculate the blocking spectrum of the target signal and non-target signal based on the best matching slope and the target slope interval, and use the blocking spectrum to construct the result of the first gain applied to the delay and sum algorithm.

[0150] Suppressing non-target sound sources, i.e., suppressing the target slope range [b m1 b m2 Suppressing external sound sources. Specifically, step S7 further includes:

[0151] Step S-700: Calculate the compensated spectrum for the two end-fire directions, obtain the blocking spectrum for the two end-fire directions based on the compensated spectrum, and obtain the first blocking spectrum AD based on the blocking spectrum for the two end-fire directions. b1 (k).

[0152] First, the compensated spectra in the two end-fire directions are calculated using the aforementioned formulas (15) and (16) (with slopes of B and -B respectively), thus obtaining the compensated spectra.

[0153] Then, based on the compensation spectrum of the two end-fire directions, the blocking spectrum of the two end-fire directions is obtained by differential calculation:

[0154]

[0155]

[0156] D b1(k) and D b2 (k) is the blocking spectrum in the end-fire direction. Calculate the amplitude of this spectrum and take the smaller value as the first blocking spectrum AD. b1 (k).

[0157] AD b1 (k)=min(|D b1 (k)|,|D b2 (k)|) (33)

[0158] Step S-701, if the best matching slope b is located in the target slope interval [b m1 b m2 Within [the specified range], the blocking spectrum D with the best matching slope b is calculated. b3 (k), and take its absolute value as the second blocking spectrum AD. b2 (k):

[0159] AD b2 (k)=|D b3 (k)| (34)

[0160] Step S-702, if the best matching slope b lies in the interval [b m1 b m2 In addition, the blocking spectrum D with the best matching slope b is calculated. b3 (k), and based on the blocking spectrum D of the best matching slope b. b3 (k) and the first blocking spectrum AD b1 (k) Update the first blocking spectrum AD b1 (k), and simultaneously re-enter the interval [b] m1 b m2 Find the best matching slope within the range and calculate the corresponding blocking spectrum amplitude as the second blocking spectrum AD. b2 (k).

[0161] Specifically, if the best matching slope b lies in the interval [b m1 b m2 In addition, the blocking spectrum D with the best matching slope b is calculated. b3 (k), then in the first blocking spectrum AD b1 (k) and spectrum D b3 (k) Take the smaller value from the absolute values ​​and update the first blocking spectrum AD. b1 (k):

[0162] AD b1 (k)=min(AD b1 (k), |D b3 (k)|) (35)

[0163] Furthermore, at this point, it is necessary to [b]m1 b m2 Within the range, re-find the best matching slope b′ and calculate the blocking spectrum D′ of the best matching slope b′. b3 (k), and then take the absolute value of each as the second blocking spectrum AD. b2 (k):

[0164] AD b2 (k)=|D′ b3 (k)| (36)

[0165] Step S-703, based on the first blocking spectrum AD b1 (k), Second blocking spectrum AD b2 (k) Construct the result of the first gain factor gain1 applied to the delay and sum algorithm.

[0166] Specifically, AD b1 (k) contains virtually no sound source information from non-target directions, AD b2 (k) contains virtually no sound source information in the target direction; the first gain factor is constructed using these two blocking spectra:

[0167]

[0168] Here, β is a number between (0,1) and close to 0, and γ is a number slightly greater than 1. Adjusting the β and γ parameters can control the suppression intensity.

[0169] Formula (37) indicates that when the sound from the target direction is less than the sound from the non-target direction, the frequency point is more likely to be the non-target sound, and it is directly suppressed; otherwise, the non-target sound is used as a reference for filtering. Applying gain1(k) to D′2(k) will yield the spectrum of the non-target direction that has been suppressed.

[0170] However, due to factors such as microphone consistency and microphone spacing, low-frequency energy from non-target directions is difficult to suppress effectively using only the methods described above. Therefore, additional suppression measures are required for low frequencies.

[0171] Optionally, the speech enhancement method for a sound source in this embodiment further includes:

[0172] Step S-8: The second gain factor is determined by comparing the spectral phase difference between the first microphone 1 and the second microphone 2 with the phase difference interval corresponding to the target slope interval, so as to suppress signals outside the target phase difference.

[0173] Since the blocking spectrum magnitudes in the target direction and the non-target direction are very similar at low frequencies, the gain obtained in step S-7 has a limited effect on suppressing low frequencies. Furthermore, because the phase difference is small at low frequencies, phase aliasing is less likely to occur. Therefore, this embodiment constructs the gain factor by directly calculating the phase difference at low frequencies.

[0174] Specifically, step S-8 further includes:

[0175] Step S-800: Calculate the spectral phase difference diffA between the second microphone 2 and the first microphone 1, and compare it with the target slope interval [b]. m1 b m2 The corresponding phase difference intervals are compared.

[0176] Specifically, the spectral phase difference diffA between the second microphone 2 and the first microphone 1 is:

[0177] diffA = angleD2 - angleD1 (38)

[0178] Step S-801, based on the spectral phase difference diffA and the slope of the target region [b m1 b m2 Determine the second gain factor to suppress signals other than the target phase difference.

[0179] If the spectral phase difference diffA is located within the slope range of the target region [b] m1 b m2 If the phase difference is outside the corresponding phase difference, suppression is performed; otherwise, no suppression is performed, thus obtaining the second gain factor.

[0180]

[0181] Where ε represents a number between (0,1) and indicates the suppression strength of low-frequency non-target signals.

[0182] Step S-802: The first gain factor and the second gain factor are applied together to the result of the delay and sum algorithm.

[0183] Specifically, gain1(k) and gain2(k) are applied together to D. out (k) can achieve a relatively ideal non-target signal suppression effect, as detailed below:

[0184] D′ out (k)=D out (k)*gain1(k)*gain2(k) (40)

[0185] As can be seen, this embodiment locates the sound source by sound source localization, determines the target location range, and then enhances the sound from the target location while suppressing the sound from non-target locations. If it is allowed to automatically find the sound source location and enhance it, this solution can be used in fields such as conference systems and common hands-free voice calls; alternatively, it can enhance the sound from a fixed direction directly without sound source localization, which can be used in fields such as directional enhancement of security cameras and headset calls.

[0186] Figure 6 The actual simulation results of this embodiment are shown: for the stage where both target speech and non-target speech exist simultaneously, the target speech spectrum becomes clearer, while non-target speech and background noise are suppressed. The actual listening experience is reduced interference and clearer target speech. For the stage with only non-target speech, both non-target speech and background noise are suppressed. This algorithm can also be used in multi-microphone scenarios, by dividing the microphones into pairs for enhancement.

[0187] Verification has shown that this method can enhance the target sound source while suppressing sounds from non-target directions, and it can also enhance speech in far-field and reverberant environments. Furthermore, this invention has the advantage of low computational cost and can easily run on common CPUs.

[0188] Example 3

[0189] Figure 7 This is a system structure diagram of a sound source localization device provided in an exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as... Figure 7 As shown, it includes:

[0190] The time-frequency conversion and polar coordinate processing unit 701 is used to convert the dual microphone signals to the frequency domain and convert the frequency domain signals into polar coordinates.

[0191] Specifically, the time-frequency conversion and polar coordinate processing unit 701 further includes

[0192] The time-frequency conversion module is used to convert the signals from two microphones into frequency domain signals.

[0193] The polar coordinate determination module is used to obtain the polar coordinate representation of the spectrum of each microphone signal.

[0194] The phase difference slope division unit 702 is used to calculate the maximum phase difference slope of the microphone signal based on the microphone spacing, and to divide the maximum phase difference slope according to the accuracy requirements.

[0195] In this embodiment, the phase difference slope division unit 702 further includes:

[0196] The phase difference slope range determination module is used to calculate the range of the phase difference slope between the two microphones based on the distance between them.

[0197] The slope division module is used to divide the phase difference slope equally according to the accuracy requirements.

[0198] The phase compensation unit 703 is used to convert the slope of each phase difference into the spectrum of the second microphone with phase compensation, and to calculate the difference matrix and the summation matrix based on the spectrum of the second microphone and the spectrum of the first microphone after compensation.

[0199] The best matching slope calculation unit 704 is used to calculate the best matching slope based on the difference matrix and the summation matrix.

[0200] The azimuth estimation unit 705 infers the delay of the signal arriving at the microphone based on the obtained best matching slope, and uses trigonometric relationships to calculate the signal azimuth as the azimuth output.

[0201] Example 4

[0202] Figure 8 This is a schematic diagram of the sound source speech enhancement device provided in the fourth exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as... Figure 8 As shown, it includes:

[0203] The time-frequency conversion and polar coordinate processing unit 801 is used to convert the dual microphone signals to the frequency domain and convert the frequency domain signals into polar coordinate form.

[0204] The phase difference slope division unit 802 is used to calculate the maximum phase difference slope of the microphone signal based on the microphone spacing, and to divide the maximum phase difference slope according to the accuracy requirements.

[0205] The phase compensation unit 803 is used to convert the slope of each phase difference into the spectrum of the second microphone with phase compensation, and to calculate the difference matrix and the summation matrix based on the spectrum of the second microphone and the spectrum of the first microphone after compensation.

[0206] The best matching slope calculation unit 804 is used to calculate the best matching slope based on the difference matrix and the summation matrix.

[0207] Since the time-frequency conversion and polar coordinate processing unit 801 and the best matching slope calculation unit 804 are the same as in Embodiment 3, they will not be described again in this embodiment.

[0208] The slope extension unit 805 is used to extend the best matching slope and combine it with the specified target sound source azimuth range to obtain the target slope range.

[0209] The spectrum compensation unit 806 is used to calculate the compensation spectrum of the second microphone 2 according to the target slope range, and to enhance the target speech by combining the spectrum signal of the first microphone 1 with the spectrum signal of the first microphone 1 using a delay and sum algorithm.

[0210] Optionally, to suppress non-target sound sources, the sound source speech enhancement device in this embodiment further includes:

[0211] The blocking spectrum and gain calculation unit 807 is used to calculate the blocking spectrum of the target signal and non-target signal based on the best matching slope and the target slope interval, and to construct the result of the first gain applied to the Delay and sum algorithm using the blocking spectrum.

[0212] Specifically, the blocking spectrum and gain calculation unit 807 further includes:

[0213] Calculate the compensated spectrum for the two end-fire directions, obtain the blocking spectrum for the two end-fire directions based on the compensated spectrum, and obtain the first blocking spectrum AD based on the blocking spectrum for the two end-fire directions. b1 (k).

[0214] If the best matching slope b is located within the target slope interval [b m1 b m2 Within [the specified range], the blocking spectrum D with the best matching slope b is calculated. b3 (k), and take its absolute value as the second blocking spectrum AD. b2 (k).

[0215] If the best matching slope b lies in the interval [b m1 b m2 In addition, the blocking spectrum D with the best matching slope b is calculated. b3 (k), and based on the blocking spectrum D of the best matching slope b. b3 (k) and the first blocking spectrum AD b1 (k) Update the first blocking spectrum AD b1 (k), and simultaneously re-enter the interval [b] m1 b m2 Find the best matching slope within the range and calculate the corresponding blocking spectrum amplitude as the second blocking spectrum AD. b2 (k).

[0216] Based on the first blocking spectrum AD b1 (k), Second blocking spectrum AD b2 (k) Construct the result of the first gain factor gain1 applied to the delay and sum algorithm.

[0217] Optionally, the sound source speech enhancement device in this embodiment further includes:

[0218] The phase difference comparison and gain calculation unit 808 is used to compare the spectral phase difference between the first microphone 1 and the second microphone 2 with the phase difference interval corresponding to the target slope interval to determine the second gain factor, so as to suppress signals other than the target phase difference.

[0219] Since the blocking spectrum magnitudes in the target direction and the non-target direction are very similar at low frequencies, the blocking spectrum and the gain calculated by the gain calculation unit 807 have limited effect on suppressing low frequencies. Furthermore, because the phase difference is small at low frequencies, phase aliasing is less likely to occur. Therefore, this embodiment constructs the gain factor by directly calculating the phase difference at low frequencies.

[0220] Specifically, the phase difference comparison and gain calculation unit 808 further includes:

[0221] Calculate the spectral phase difference diffA between the second microphone 2 and the first microphone 1, and compare it with the target slope interval [b]. m1 b m2 The corresponding phase difference intervals are compared.

[0222] Based on the spectral phase difference diffA and the slope of the target region [b] m1 b m2 Determine the second gain factor to suppress signals other than the target phase difference.

[0223] The first gain factor and the second gain factor are applied together to the result of the delay and sum algorithm.

[0224] Exemplary electronic devices

[0225] Figure 9 This is the structure of an electronic device provided in an exemplary embodiment of the present invention. The electronic device may be either or both of a first device and a second device, or a standalone device independent of them, which may communicate with the first device and the second device to receive acquired input signals from them. Figure 9 A block diagram of an electronic device according to an embodiment of the present disclosure is shown. Figure 9 As shown, the electronic device includes one or more processors 91 and memory 92.

[0226] The processor 91 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0227] The memory 92 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 91 may execute the program instructions to implement the sound source localization method and / or other desired functions of the software program of the various embodiments of this disclosure described above. In one example, the electronic device may further include an input device 93 and an output device 94, these components being interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0228] In addition, the input device 93 may also include, for example, a keyboard, a mouse, etc.

[0229] The output device 94 can output various information to the outside. The output device 94 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0230] Of course, for the sake of simplicity, Figure 9 Only some of the components of the electronic device relevant to this disclosure are shown, omitting components such as buses, input / output interfaces, etc. In addition, the electronic device may include any other suitable components depending on the specific application.

[0231] Exemplary computer program products and computer-readable storage media

[0232] In addition to the methods and apparatus described above, embodiments of this disclosure may also be computer program products comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the sound source localization methods according to various embodiments of this disclosure as described in the "Exemplary Methods" section of this specification.

[0233] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this disclosure. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on a user's computing device, partially on a user's computing device, as a standalone software package, partially on a user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0234] Furthermore, embodiments of this disclosure may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps in the sound source localization methods according to various embodiments of this disclosure as described in the "Exemplary Methods" section above.

[0235] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0236] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0237] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For system embodiments, since they largely correspond to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0238] The block diagrams of devices, apparatuses, devices, and systems disclosed herein are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0239] The methods and apparatus of this disclosure may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the methods is for illustrative purposes only, and the steps of the methods of this disclosure are not limited to the order specifically described above unless otherwise specifically stated. Furthermore, in some embodiments, this disclosure may also be implemented as a program recorded on a recording medium, the program including machine-readable instructions for implementing the methods according to this disclosure. Thus, this disclosure also covers recording media storing programs for performing the methods according to this disclosure.

[0240] It should also be noted that in the apparatus, devices, and methods of this disclosure, the components or steps are decomposable and / or recombinable. Such decomposition and / or recombination should be considered equivalent to the present disclosure. The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0241] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A sound source localization method, comprising the following steps: The dual-microphone signals are converted to the frequency domain, and the frequency domain signals are then converted to polar coordinates. The slope range of the microphone signal phase difference is calculated based on the microphone spacing and sound speed, and the phase difference slope is divided according to the accuracy requirements. The slopes of each phase difference are converted into the spectrum of the second microphone after phase compensation, and the difference matrix and summation matrix are calculated based on the spectrum of the second microphone after compensation and the spectrum of the first microphone. The best-matching slope is calculated based on the difference matrix and the summation matrix. Based on the obtained best-matching slope, the delay of the signal arriving at the microphone is deduced, and the signal orientation is determined using trigonometric relationships, which is then used as the orientation output.

2. The sound source localization method as described in claim 1, characterized in that, In the step of calculating the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and dividing the phase difference slope according to the accuracy requirements, the maximum phase difference slope of the first microphone and the second microphone is calculated based on the microphone spacing and sound speed, the phase difference slope range is determined based on the maximum phase difference slope, and the phase difference slope is divided equally according to the accuracy requirements.

3. The sound source localization method as described in claim 2, characterized in that, The step of converting the slopes of each phase difference into the phase-compensated spectrum of the second microphone, and calculating the difference matrix and summation matrix based on the compensated second microphone spectrum and the first microphone spectrum, includes: The phase differences represented by each slope are sequentially compensated to the second microphone, and the compensated spectrum of the second microphone is recovered from the compensated phase. Subtracting the phase-compensated spectrum of the second microphone from the spectrum of the first microphone yields the MD. m (k) Add the phase compensation spectrum of the second microphone to the spectrum of the first microphone to obtain PD. m (k), and finally MD respectively m (k) and PD m (k) Summing the absolute values ​​of the frequency points yields the difference matrix and the summation matrix.

4. The sound source localization method as described in claim 3, characterized in that, In the step of calculating the best matching slope based on the difference matrix and the summation matrix, the difference matrix and the summation matrix are summed according to the number of frames to obtain two vectors. Then, the two vectors are compared, and the phase difference slope corresponding to the smallest ratio is selected as the best matching slope.

5. A method for enhancing speech from a sound source, comprising the following steps: The dual-microphone signals are converted to the frequency domain, and the frequency domain signals are then converted to polar coordinates. The slope range of the microphone signal phase difference is calculated based on the microphone spacing and sound speed, and the phase difference slope is divided according to the accuracy requirements. The slopes of each phase difference are converted into the spectrum of the second microphone after phase compensation, and the difference matrix and summation matrix are calculated based on the spectrum of the second microphone after compensation and the spectrum of the first microphone. The best-matching slope is calculated based on the difference matrix and the summation matrix. The best matching slope is extended and combined with the specified target sound source azimuth range to obtain the target slope range; The compensation spectrum of the second microphone is calculated based on the target slope range, and the target speech is enhanced by using a delay accumulation algorithm together with the spectrum signal of the first microphone.

6. The speech enhancement method for a sound source as described in claim 5, characterized in that, The method further includes: The blocking spectrum of the target signal and the non-target signal is calculated based on the best matching slope and the target slope interval, and the blocking spectrum is used to construct the result of the first gain applied to the delay accumulation algorithm.

7. The speech enhancement method for a sound source as described in claim 6, characterized in that, The step of calculating the blocking spectrum of the target signal and non-target signal based on the best matching slope and the target slope interval, and using the blocking spectrum to construct the result of the first gain applied to the delay accumulation algorithm, further includes: Calculate the compensated spectrum for the two end-fire directions, and obtain the blocking spectrum for the two end-fire directions based on the compensated spectrum. Obtain the first blocking spectrum based on the blocking spectrum for the two end-fire directions. If the best matching slope is within the target slope range, then the blocking spectrum of the best matching slope is calculated, and its absolute value is taken as the second blocking spectrum. If the best matching slope is outside the target slope interval, the blocking spectrum of the best matching slope is calculated, and the first blocking spectrum is updated based on the blocking spectrum of the best matching slope b and the first blocking spectrum. At the same time, the best matching slope is searched again within the target slope interval, and the corresponding blocking spectrum amplitude is calculated as the second blocking spectrum. The result of constructing a first gain factor based on the first blocking spectrum and the second blocking spectrum and applying it to the delay accumulation algorithm.

8. The speech enhancement method for a sound source as described in claim 7, characterized in that, The method further includes: A second gain factor is determined by comparing the spectral phase difference between the first and second microphones with the phase difference interval corresponding to the target slope interval, so as to suppress signals outside the target phase difference.

9. A sound source localization device, comprising: The time-frequency conversion and polar coordinate processing unit is used to convert the dual microphone signals to the frequency domain and convert the frequency domain signals into polar coordinate form; The phase difference slope division unit is used to calculate the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and to divide the phase difference slope according to the accuracy requirements. The phase compensation unit is used to convert the slopes of each phase difference into the spectrum of the second microphone with phase compensation, and to calculate the difference matrix and the summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone. The optimal matching slope calculation unit is used to calculate the optimal matching slope based on the difference matrix and the summation matrix. The azimuth estimation unit infers the delay of the signal arriving at the microphone based on the obtained best-matching slope, and uses trigonometric relationships to calculate the signal azimuth as the azimuth output.

10. A sound source speech enhancement device, comprising: The time-frequency conversion and polar coordinate processing unit is used to convert the dual microphone signals to the frequency domain and convert the frequency domain signals into polar coordinate form; The phase difference slope division unit is used to calculate the slope range of the microphone signal phase difference based on the microphone spacing and sound speed, and to divide the phase difference slope according to the accuracy requirements. The phase compensation unit is used to convert the slopes of each phase difference into the spectrum of the second microphone with phase compensation, and to calculate the difference matrix and the summation matrix based on the compensated spectrum of the second microphone and the spectrum of the first microphone. The optimal matching slope calculation unit is used to calculate the optimal matching slope based on the difference matrix and the summation matrix. The slope expansion unit is used to expand the best matching slope and combine it with the specified target sound source azimuth range to obtain the target slope range; The spectrum compensation unit is used to calculate the compensation spectrum of the second microphone based on the target slope range, and to enhance the target speech by combining it with the spectrum signal of the first microphone using a delay accumulation algorithm.

Images