An adaptive directional voice-following dual-microphone noise reduction method and system

Abstract

The present invention discloses an adaptive directional following voice dual-microphone noise reduction method and system. Firstly, a first original signal and a second original signal collected by a first microphone and a second microphone are framed, and a DC component elimination module is used to remove the DC offset component and obtain the first microphone signal after DC removal and the second microphone signal after DC removal. Then, the first microphone signal after DC removal is sent to a voice detection module for processing and the probability of voice occurrence in the current environment is obtained. The present invention realizes the function of following the voice orientation and direction noise reduction, and can adaptively detect and follow the voice in the environment to amplify the signal amplitude in the voice direction and suppress noise in other directions, thereby improving the signal-to-noise ratio and clarity. It also has the directionality increase characteristic that can be applied to various application scenarios, has good application prospects, and is suitable for being widely promoted and used.

Description

Technical Field

[0001] The present invention relates to the technical field of speech signal processing, and in particular to an adaptive directional follow-up speech dual-microphone noise reduction method and system. Background Art

[0002] As the number of people suffering from hearing impairment increases, the popularity of hearing aids is also increasing. More and more hearing-impaired patients are willing and able to choose to improve their hearing conditions through hearing aids. Therefore, hearing aid technology has also ushered in new developments. Although traditional analog hearing aids are low in cost, they also have a series of problems such as low accuracy, relatively simple functions, and signal distortion. Digital hearing aids have become the choice of more and more people without significantly increasing costs, relying on the advantages of programmability, configurability, rich algorithms, low distortion, and high accuracy. Therefore, research on digital hearing aid algorithms is not only necessary, but also has far-reaching impacts.

[0003] The advantage of multiple microphones over a single microphone is that they can provide more spatial information, realize more complex functions, and provide better listening effects. For example, the multi-microphone structure can realize the function of directional sound reception, effectively suppress signals outside the target direction, and significantly improve the signal-to-noise ratio and clarity of the signal.

[0004] At present, most digital hearing aids are based on a single microphone structure. Since dual microphones can extract more spatial and directional information, the dual microphone structure can effectively improve the signal-to-noise ratio and clarity of the hearing aid in a noisy environment, thereby realizing more complex directionality and noise reduction algorithms. Moreover, the directional algorithms of dual-microphone hearing aids are mostly fixed directionality, and their suppression and enhancement directions are fixed, while in actual application scenarios, target information such as speech is often in all directions; therefore, it is necessary to design an adaptive directional following speech dual microphone noise reduction method and system. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and to better and effectively solve the problem that the directional algorithms of current dual-microphone hearing aids are mostly fixed directional, and their suppression and enhancement directions are fixed, while in actual application scenarios, target information such as speech is often in various directions. An adaptive directional following speech dual-microphone noise reduction method and system are provided, which realizes the function of following the speech orientation and direction noise reduction, and can adaptively detect and follow the speech in the environment to amplify the signal amplitude in the speech direction and suppress noise in other directions, thereby improving the signal-to-noise ratio and clarity, and also has the directionality increase characteristics that can be applied to various application scenarios, and has good application prospects.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] An adaptive directional follow-up voice dual-microphone noise reduction method and system, comprising the following steps:

[0008] Step A: collecting the first original signal from the first microphone and the second microphone and the second original signal After framing, the DC component elimination module is used to remove the DC offset component to obtain the first microphone signal after DC removal. and the second microphone signal after DC removal ;

[0009] Step B: Remove the DC from the first microphone signal Send it to the speech detection module for processing to obtain the probability of speech occurrence in the current environment ;

[0010] Step C, the first microphone signal after DC removal Send it to the energy detection module for processing to obtain the energy level in the current environment ;

[0011] Step D, based on the obtained probability of speech occurrence in the current environment and the energy level of the current environment , use the low-precision azimuth calculation module or the high-precision azimuth calculation module to calculate and obtain the sound source azimuth ;

[0012] Step E: remove the DC first microphone signal , Second microphone signal after DC removal and sound source location The signal is input to the adaptive directional signal synthesis module for processing and output as a synthesized signal.

[0013] The aforementioned adaptive directional following voice dual microphone noise reduction method, step A, the first original signal collected by the first microphone and the second microphone and the second original signal After framing, the DC component elimination module is used to remove the DC offset component to obtain the first microphone signal after DC removal. and the second microphone signal after DC removal , where the DC component elimination module uses a high-pass filter.

[0014] In the aforementioned adaptive directional following voice dual microphone noise reduction method, step B, the first microphone signal after DC removal is Send it to the speech detection module for processing to obtain the probability of speech occurrence in the current environment , the specific steps are as follows,

[0015] Step B1, calculate the signal energy of the mth frame in the current environment , as shown in formula (1),

[0016] (1)

[0017] in, is the number of samples per frame, is the square of the sampling value of the nth sample signal;

[0018] Step B2, calculating the first power statistic , as shown in formula (2),

[0019] (2)

[0020] in, and is a constant coefficient;

[0021] Step B3, calculating the second power statistic , as shown in formula (3),

[0022] (3)

[0023] in, The number of historical frames taken for history, is the sampling point number;

[0024] Step B4, calculate the Speech probability of the frame , as shown in formula (4),

[0025] (4);

[0026] Step B5: speech probability Smoothing is performed to obtain The probability of speech occurrence in the current frame environment , the specific calculation method is shown in the following formula (5):

[0027] (5)

[0028] in, and is the smoothing coefficient.

[0029] In the aforementioned adaptive directional following voice dual microphone noise reduction method, step C, the first microphone signal after DC removal is Send it to the energy detection module for processing to obtain the energy level in the current environment , as shown in formula (6),

[0030] (6).

[0031] In the aforementioned adaptive directional following voice dual microphone noise reduction method, step D, based on the obtained voice occurrence probability in the current environment and the energy level of the current environment , using low-precision azimuth measurement module Or high-precision position measurement module Calculate and get the direction of the sound source , the specific steps are as follows,

[0032] Step D1: Building a low-precision position measurement module The implementation method is shown in formula (7):

[0033] (7)

[0034] in, and is a constant;

[0035] Step D2: construct a high-precision position estimation unit The implementation method is shown in formula (8):

[0036] (8);

[0037] Step D3: Use low-precision azimuth measurement module Or high-precision position measurement module Calculate and get the direction of the sound source ,

[0038] (9)

[0039] in, is the speech probability activation threshold, is the sound pressure level threshold; if and When , it means that there is voice in the current signal, and the high-precision direction measurement module is used Otherwise, the low-precision position calculation module is used. Make calculations.

[0040] In the aforementioned adaptive directional following voice dual microphone noise reduction method, step E, the first microphone signal after DC removal is , Second microphone signal after DC removal and sound source location Input to the adaptive directional signal synthesis module and process it, and output the synthesized signal. The specific steps are as follows:

[0041] Step E1: remove the DC first microphone signal , Second microphone signal after DC removal Delay filter Delay and get the first microphone signal after delay and the delayed second microphone signal , as shown in formula (10) and formula (11),

[0042] (10)

[0043] (11);

[0044] Step E2, according to the first microphone signal after DC removal , the second microphone signal after DC removal, the first microphone signal after delay , Delayed second microphone signal and sound source location Perform synthesis to obtain a synthetic signal , as shown in formula (12),

[0045] (12).

[0046] The aforementioned adaptive directional follow-up voice dual microphone noise reduction method, the delay filter in step E1 A non-recursive finite impulse response filter is used, and the second coefficient calculation steps of the non-recursive finite impulse response filter are as follows:

[0047] Step E11, according to the end-to-end distance of the dual microphones Calculate fixed signal delay time , as shown in formula (13),

[0048] (13)

[0049] in, is the speed of sound in air, is the signal sampling rate;

[0050] Step E12, set the filter length to , then the first delay control vector As shown in formula (14),

[0051] (14);

[0052] Step E13, calculating the second delay control vector As shown in formula (15),

[0053] (15);

[0054] Step E14, calculating the first coefficient of the non-recursive finite impulse response filter As shown in formula (16),

[0055] (16)

[0056] Step E15, the first coefficient of the non-recursive finite impulse response filter Windowing is performed to obtain the second coefficient of the non-recursive finite impulse response filter .

[0057] An adaptive directional follow-up voice dual-microphone noise reduction system includes a DC removal unit, a voice detection unit, an energy detection unit, a sound source orientation measurement unit and a signal synthesis unit. The DC removal unit is used to combine the first original signal collected by the first microphone and the second microphone. and the second original signal After framing, the DC component elimination module is used to remove the DC offset component and obtain the first microphone signal after DC removal and the second microphone signal after DC removal ;

[0058] The voice detection unit is used to detect the first microphone signal after DC removal. Send it to the speech detection module for processing and obtain the probability of speech occurrence in the current environment ;

[0059] The energy detection unit is used to detect the first microphone signal after the DC is removed. Send it to the energy detection module for processing and obtain the energy level in the current environment ;

[0060] The sound source location estimation unit is used to obtain the probability of speech occurrence in the current environment based on the obtained and the energy level of the current environment , and then use the low-precision azimuth calculation module or the high-precision azimuth calculation module to calculate and obtain the sound source azimuth ;

[0061] The signal synthesis unit is used to remove the DC first microphone signal , Second microphone signal after DC removal and sound source location The signal is input to the adaptive directional signal synthesis module for processing and outputs the synthesized signal.

[0062] The beneficial effects of the present invention are as follows: an adaptive directional following voice dual-microphone noise reduction method and system of the present invention first divides the first original signal and the second original signal collected by the first microphone and the second microphone into frames, and then respectively uses a DC component elimination module to remove the DC offset component and obtains the first microphone signal after DC removal and the second microphone signal after DC removal, then sends the first microphone signal after DC removal to a voice detection module for processing and obtains the probability of speech occurrence in the current environment, then sends the first microphone signal after DC removal to an energy detection module for processing and obtains the energy level in the current environment, and then based on the obtained speech occurrence in the current environment The probability and the energy level in the current environment are calculated using a low-precision direction measurement module or a high-precision direction measurement module to obtain the direction of the sound source, and then the first microphone signal after DC removal, the second microphone signal after DC removal and the sound source direction are input into the adaptive directional signal synthesis module for processing and output of the synthesized signal. The present invention effectively realizes the function of following the speech direction and directional noise reduction, and can adaptively detect and follow the speech in the environment to amplify the signal amplitude in the direction of the speech and suppress noise in other directions, thereby improving the signal-to-noise ratio and clarity, and also has the directionality increase characteristic that can be applied to various application scenarios, and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is an overall flow chart of an adaptive directional follow-up voice dual-microphone noise reduction method of the present invention;

[0064] Figure 2 It is a schematic diagram of the working principle of an adaptive directional voice-following dual-microphone noise reduction system of the present invention;

[0065] Figure 3 1 is a comparison diagram of input and output effects of a dual-microphone adaptive directional noise reduction method following speech in an embodiment of the present invention;

[0066] Figure 4 This is a diagram showing the effect of a conventional dual-microphone delayed addition algorithm in an embodiment of the present invention;

[0067] Figure 5 3 is a comparison diagram between the original input signal and the speech probability in an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The present invention will be further described below in conjunction with the accompanying drawings.

[0069] like Figure 1 As shown, an adaptive directional following voice dual microphone noise reduction method and system of the present invention comprises the following steps:

[0070] Step A: collecting the first original signal from the first microphone and the second microphone and the second original signal After framing, the DC component elimination module is used to remove the DC offset component to obtain the first microphone signal after DC removal. and the second microphone signal after DC removal , where the DC component elimination module uses a high-pass filter.

[0071] Step B: Remove the DC from the first microphone signal Send it to the speech detection module for processing to obtain the probability of speech occurrence in the current environment , the specific steps are as follows,

[0072] Step B1, calculate the signal energy of the mth frame in the current environment , as shown in formula (1),

[0073] (1)

[0074] in, is the number of samples per frame, is the square of the sampling value of the nth sample signal;

[0075] Step B2, calculating the first power statistic , as shown in formula (2),

[0076] (2)

[0077] in, and is a constant coefficient;

[0078] Step B3, calculating the second power statistic , as shown in formula (3),

[0079] (3)

[0080] in, The number of historical frames taken for history, is the sampling point number;

[0081] Step B4, calculate the Speech probability of the frame , as shown in formula (4),

[0082] (4);

[0083] Step B5: speech probability Smoothing is performed to obtain The probability of speech occurrence in the current frame environment , the specific calculation method is shown in the following formula (5):

[0084] (5)

[0085] in, and is the smoothing coefficient.

[0086] Step C, the first microphone signal after DC removal Send it to the energy detection module for processing to obtain the energy level in the current environment , as shown in formula (6),

[0087] (6).

[0088] Step D, based on the obtained probability of speech occurrence in the current environment and the energy level of the current environment , using low-precision azimuth measurement module Or high-precision position measurement module Calculate and get the direction of the sound source , the specific steps are as follows,

[0089] Step D1: Building a low-precision position measurement module The implementation method is shown in formula (7):

[0090] (7)

[0091] in, and is a constant;

[0092] Step D2: Building a high-precision position measurement module The implementation method is shown in formula (8):

[0093] (8);

[0094] Step D3: Use low-precision azimuth measurement module Or high-precision position measurement module Calculate and get the direction of the sound source ,

[0095] (9)

[0096] in, is the speech probability activation threshold, is the sound pressure level threshold; if and When , it means that there is voice in the current signal, and the high-precision direction measurement module is used Otherwise, the low-precision position calculation module is used. Make calculations.

[0097] Step E: remove the DC first microphone signal , Second microphone signal after DC removal and sound source location Input to the adaptive directional signal synthesis module and process it, and output the synthesized signal. The specific steps are as follows:

[0098] Step E1: remove the DC first microphone signal , Second microphone signal after DC removal Delay filter Delay and get the first microphone signal after delay and the delayed second microphone signal , as shown in formula (10) and formula (11),

[0099] (10)

[0100] (11);

[0101] Step E2, according to the first microphone signal after DC removal , the second microphone signal after DC removal, the first microphone signal after delay , Delayed second microphone signal and sound source location Perform synthesis to obtain a synthetic signal , as shown in formula (12),

[0102] (12).

[0103] Delay filter in step E1 A non-recursive finite impulse response filter is used, and the second coefficient calculation steps of the non-recursive finite impulse response filter are as follows:

[0104] Step E11, according to the end-to-end distance of the dual microphones Calculate fixed signal delay time , as shown in formula (13),

[0105] (13)

[0106] in, is the speed of sound in air, is the signal sampling rate;

[0107] Step E12, set the filter length to , then the first delay control vector As shown in formula (14),

[0108] (14);

[0109] Step E13, calculating the second delay control vector As shown in formula (15),

[0110] (15);

[0111] Step E14, calculating the first coefficient of the non-recursive finite impulse response filter As shown in formula (16),

[0112] (16)

[0113] Step E15, the first coefficient of the non-recursive finite impulse response filter Windowing is performed to obtain the second coefficient of the non-recursive finite impulse response filter .

[0114] like Figure 2 As shown, an adaptive directional follow-up voice dual-microphone noise reduction system includes a DC removal unit, a voice detection unit, an energy detection unit, a sound source direction estimation unit and a signal synthesis unit. The DC removal unit is used to combine the first original signal collected by the first microphone and the second microphone. and the second original signal After framing, the DC component elimination module is used to remove the DC offset component and obtain the first microphone signal after DC removal and the second microphone signal after DC removal ;

[0115] The voice detection unit is used to detect the first microphone signal after DC removal. Send it to the speech detection module for processing and obtain the probability of speech occurrence in the current environment ;

[0116] The energy detection unit is used to detect the first microphone signal after the DC is removed. Send it to the energy detection module for processing and obtain the energy level in the current environment ;

[0117] The sound source location estimation unit is used to obtain the probability of speech occurrence in the current environment based on the obtained and the energy level of the current environment , and then use the low-precision azimuth calculation module or the high-precision azimuth calculation module to calculate and obtain the sound source azimuth ;

[0118] The signal synthesis unit is used to remove the DC first microphone signal , Second microphone signal after DC removal and sound source location The signal is input to the adaptive directional signal synthesis module for processing and outputs the synthesized signal.

[0119] In order to better illustrate the use effect of the present invention, a specific embodiment of using the present invention to reduce noise for dual microphones is introduced below.

[0120] like Figure 3 As shown, it is a comparison diagram of the input and output effects of the dual-microphone adaptive directional noise reduction method following the speech in an embodiment of the present invention. The input signal of the left channel is the sound signal of the outdoor noisy environment, and the output signal of the right channel is the output signal of the algorithm; it can be observed from the input and output comparison signals that in the outdoor noisy environment, the collected input signal SNR is low, and the output signal can effectively follow and extract the speech signal in the noisy environment, effectively suppress the noise in the environment, improve the SNR, and improve the clarity and quality of the signal;

[0121] like Figure 4 As shown in the figure, it is an effect diagram of a conventional dual-microphone delay addition algorithm in an embodiment of the present invention. The input signal of the left channel is a sound signal of an outdoor noisy environment, and the output signal of the right channel is the output signal of the delay addition algorithm. The delay addition algorithm delays the second microphone signal and adds it to the first microphone signal and then averages it. By comparing Figure 4 The output signal Figure 3 From the output signal and input signal, it can be seen that the effect of the present invention is better than the traditional method: it can increase the SNR more and significantly improve the listening effect.

[0122] like Figure 5 As shown, the original input signal and the speech probability in the embodiment of the present invention are Comparison chart, the input signal of the left channel is the sound signal of the outdoor noisy environment, and the right channel is the speech probability of the corresponding frame calculated based on the input signal The larger the value, the higher the probability of speech. By comparison, it can be seen that the system can more accurately reflect the probability of speech in a noisy environment.

[0123] In summary, an adaptive directional following voice dual-microphone noise reduction method and system of the present invention first divides the first original signal and the second original signal collected by the first microphone and the second microphone into frames, and then uses a DC component elimination module to remove the DC offset component and obtain the first microphone signal after DC removal and the second microphone signal after DC removal, then sends the first microphone signal after DC removal to the voice detection module for processing and obtains the probability of speech occurrence in the current environment, and then sends the first microphone signal after DC removal to the energy detection module for processing and obtains the energy level in the current environment, and then uses a low-precision azimuth measurement module or a high-precision azimuth measurement module based on the obtained probability of speech occurrence in the current environment and the energy level in the current environment to measure and obtain the sound source azimuth, and then the first microphone signal after DC removal, the second microphone signal after DC removal and the sound source azimuth are input into the adaptive directional signal synthesis module for processing and output the synthesized signal, which effectively realizes the function of following the voice azimuth and directional noise reduction of the present invention, and can adaptively detect and follow the voice in the environment to amplify the signal amplitude in the voice direction and suppress noise in other directions, thereby improving the signal-to-noise ratio and clarity, and also has a directional increase characteristic that can be applied to various application scenarios.

[0124] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An adaptive directional follow-up voice dual-microphone noise reduction method, characterized in that: The following steps are included: Step A: After the first original signal x1(m) and the second original signal x1(m) collected by the first microphone and the second microphone are divided into frames, a DC component elimination module is used to remove the DC offset component to obtain the first microphone signal x1(m) after DC removal. 11 (m) and the second microphone signal x after DC removal 21 (m); Step B: remove the DC first microphone signal x 11 (m) is sent to the speech detection module for processing to obtain the probability v of speech occurrence in the current environment. The specific steps are as follows: Step B1, calculate the signal energy S of the mth frame in the current environment 2 (m), as shown in formula (1), in, N is the number of samples per frame, is the square of the sampling value of the nth sample signal; Step B2, calculate the first power statistic E, as shown in formula (2), E=c1·S 2 (m)+c2·S 2 (m-1)(2) Among them, c1 and c2 are constant coefficients; Step B3, calculate the second power statistic e(m), as shown in formula (3), Among them, N1 is the number of historical frames taken by history, and n is the sampling point number; Step B4, calculate the speech probability v1 of the mth frame, as shown in formula (4), Step B5, smoothing the speech probability v1 to obtain the speech occurrence probability v(m) in the current environment of the mth frame, the specific calculation method is shown in the following formula (5): v(m)=a1·v1(m)+a2·v(m-1)(5) Among them, a1 and a2 are smoothing coefficients; Step C: remove the DC first microphone signal x 11 (m) is sent to the energy detection module for processing to obtain the energy level s in the current environment; Step D, based on the obtained speech occurrence probability v and energy level s in the current environment, a low-precision orientation measurement module or a high-precision orientation measurement module is used to perform measurement to obtain the sound source orientation d(m). The specific steps are as follows: Step D1, construct a low-precision azimuth measurement module Cal_Low(). The implementation method is shown in formula (7): d(m)=b1·d(m-1)+b2·d(m-2) (7) Among them, b1 and b2 are constants; Step D2, construct a high-precision azimuth measurement module Cal_High(), and the implementation method is as shown in formula (8): Step D3, using the low-precision azimuth calculation module Cal_Low() or the high-precision azimuth calculation module Cal_High() to calculate and obtain the sound source azimuth d(m), Among them, v is the speech probability activation threshold, th SPL is the sound pressure level threshold; if v>th v And s>th SPL When , it means that there is voice in the current signal. At this time, the high-precision direction calculation module Cal_High() is used for calculation. Otherwise, the low-precision direction calculation module Cal_Low() is used for calculation. Step E: remove the DC first microphone signal x 11 (m), the second microphone signal x after DC removal 21 (m) and the sound source direction d(m) are input to the adaptive directional signal synthesis module and processed to output the synthesized signal. The specific steps are as follows: Step E1: remove the DC first microphone signal x 11 (m), the second microphone signal x after DC removal 21 (m) is delayed by the delay filter delay() to obtain the delayed first microphone signal x 11 _delay(m) and the delayed second microphone signal x 21 _delay(m), as shown in formula (10) and formula (11), x 11 _delay(m)=delay(x 11 )(10) x 21 _delay(m)=delay(x 21 )(11); Step E2, according to the first microphone signal x after DC removal 11 (m), the second microphone signal after DC removal, the first microphone signal after delay x 11 _delay(m), delayed second microphone signal x 21 _delay(m) and the sound source direction d(m) are synthesized to obtain the synthetic signal x out (m), as shown in formula (12), x out (m)=[x 11 (m)-x 21 _delay(m)]-d(m)·[x 21 (m)-x 11 _delay(m)](12)。 2. The adaptive directional follow-up voice dual-microphone noise reduction method according to claim 1, characterized in that: Step A: After the first original signal x1(m) and the second original signal x1(m) collected by the first microphone and the second microphone are divided into frames, a DC component elimination module is used to remove the DC offset component to obtain the first microphone signal x1(m) after DC removal. 11 (m) and the second microphone signal x after DC removal 21 (m), where the DC component elimination module uses a high-pass filter.

3. The adaptive directional voice-following dual-microphone noise reduction method according to claim 2, characterized in that: Step C: remove the DC first microphone signal x 11 (m) is sent to the energy detection module for processing to obtain the energy level s in the current environment, as shown in formula (6):

4. The adaptive directional follow-up voice dual-microphone noise reduction method according to claim 3, characterized in that: The delay filter delay() in step E1 adopts a non-recursive finite impulse response filter, and the second coefficient calculation step of the non-recursive finite impulse response filter is as follows: Step E11, according to the end-to-end distance d of the two microphones end_fire Calculate the fixed signal delay time Delay, as shown in formula (13), Where c is the speed of sound in air, F s is the signal sampling rate; Step E12, assuming the filter length is FL, the first delay control vector Z0 is as shown in formula (14), Z0=0...FL(14); Step E13, calculate the second delay control vector Z as shown in formula (15), Z = Z0 - Delay (15); Step E14, calculate the first coefficient Filter_weight_0 of the non-recursive finite impulse response filter as shown in formula (16), Step E15 , windowing the first coefficient Filter_weight_0 of the non-recursive finite impulse response filter to obtain the second coefficient Filter_weight of the non-recursive finite impulse response filter.

5. An adaptive directional voice-following dual-microphone noise reduction system, wherein the noise reduction process of the dual-microphone noise reduction system is based on the dual-microphone noise reduction method according to any one of claims 1 to 4, characterized in that: The invention comprises a DC removal unit, a speech detection unit, an energy detection unit, a sound source orientation measurement unit and a signal synthesis unit. The DC removal unit is used to divide the first original signal x1(m) and the second original signal x1(m) collected by the first microphone and the second microphone into frames and then use a DC component elimination module to remove the DC offset component and obtain the first microphone signal x1(m) after DC removal. 11 (m) and the second microphone signal x after DC removal 21 (m); The speech detection unit is used to remove the DC first microphone signal x 11 (m) The speech detection module is sent to the speech detection module for processing and the probability v of speech occurrence in the current environment is obtained; The energy detection unit is used to detect the first microphone signal x after the DC is removed. 11 (m) is sent to the energy detection module for processing and the energy level s in the current environment is obtained; The sound source direction calculation unit is used to calculate and obtain the sound source direction d(m) by using a low-precision direction calculation module or a high-precision direction calculation module based on the obtained speech occurrence probability v and energy level s in the current environment; The signal synthesis unit is used to synthesize the first microphone signal x after DC removal 11 (m), the second microphone signal x after DC removal 21 The sound source position d(m) and the sound source direction d(m) are input to the adaptive directional signal synthesis module for processing and output as a synthesized signal.

Images