A low-complexity and high-resolution howling detection and suppression method and system for digital hearing aids

Abstract

The present invention discloses a low-complexity, high-resolution howling detection and suppression method and system for a digital hearing aid. First, the sound signal of the surrounding environment is collected and converted into a digital signal, and then the digital signal is preprocessed to obtain a preprocessed signal. Then, the preprocessed signal is decomposed and processed by a multi-channel filter group and a multi-channel signal is output. Then, according to the energy distribution of the multi-channel signal, whether there is howling in the current environmental sound signal is detected and a howling detection result is obtained. The present invention realizes the function of reducing the calculation complexity and improving the detection accuracy through multi-level processing and gradual refinement, and the multi-channel decomposition module and the filtering module in the digital hearing aid can be deployed and run on a low-power and low-computing power platform with extremely small resource overhead, and the preset weight adjustment through the weight adjustment mechanism ensures that the system can stably and accurately detect and suppress howling in practical applications, and is suitable for being widely promoted and used.

Description

Technical Field

[0001] The present invention relates to the technical field of hearing aid equipment, and in particular to a low-complexity and high-resolution howling detection and suppression method and system for digital hearing aids. Background Art

[0002] Hearing aids are an important hearing aid device and are widely used by patients with different degrees of hearing loss. Hearing aids usually include a microphone for collecting sound signals, a signal processor for performing loudness compensation or more algorithm processing on the signals collected by the microphone, and a speaker for outputting sound.

[0003] However, in actual use, due to the small size of the hearing aid, its microphone can easily receive the amplified signal output by the speaker, causing the hearing aid to produce acoustic feedback problems and cause howling, which not only affects the user's comfort, but may also damage the user's hearing health. Currently, when using fast Fourier transform for howling detection, sufficient frequency resolution is required to improve the detection accuracy, and in the signal processing process of the digital hearing aid, computing power is also required to run the hearing aid's key hearing compensation and noise suppression algorithms, which directly increases the length of the fast Fourier transform, resulting in excessive computational complexity and storage space resource consumption. Therefore, it is necessary to design a low-complexity, high-resolution howling detection and suppression method and system for digital hearing aids. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and to better and effectively solve the problem that sufficient frequency resolution is required to improve the detection accuracy when using fast Fourier transform for howling detection. In the signal processing process of the digital hearing aid, computing power is also required to run the key hearing compensation and noise suppression algorithms of the hearing aid, which directly increases the length of the fast Fourier transform, resulting in excessive consumption of computing power and storage space resources. A low-complexity, high-resolution howling detection and suppression method and system for digital hearing aids are provided, which realizes the function of reducing the computational complexity and improving the detection accuracy through multi-level processing and gradual refinement, and can be deployed and operated on a low-power and low-computing power platform with extremely small resource overhead through the multi-channel decomposition module and the filtering module in the digital hearing aid, and the preset weight adjustment through the weight adjustment mechanism ensures that the system can stably and accurately detect and suppress howling in practical applications.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A low-complexity and high-resolution howling detection and suppression method and system for a digital hearing aid, comprising the following steps:

[0007] Step A, collecting sound signals from the surrounding environment and converting them into digital signals, and then preprocessing the digital signals to obtain preprocessed signals;

[0008] Step B, decomposing the preprocessed signal through a multi-channel filter bank and outputting a multi-channel signal;

[0009] Step C, detecting whether there is howling in the current ambient sound signal according to the energy distribution of the multi-channel signal, and obtaining a howling detection result;

[0010] Step D, using the host computer to determine whether the howling detection result is accurate, if the howling detection result is wrong, then automatically adjust the weight, if the howling detection result is correct, then close the weight adjustment process;

[0011] Step E: Calculate feedback gain based on the howling detection result and complete the howling suppression operation.

[0012] The aforementioned low-complexity and high-resolution howling detection and suppression method for digital hearing aids, step A, collects sound signals from the surrounding environment and converts them into digital signals, and then preprocesses the digital signals to obtain preprocessed signals, wherein the preprocessing of the digital signals and obtaining the preprocessed signals specifically involves statistically analyzing the short-time energy I of the filtered broadband digital signal, and if the short-time energy I is greater than the set energy threshold Th, the signal gain is reduced during the acquisition and filtering process.

[0013] The low-complexity and high-resolution howling detection and suppression method for digital hearing aids mentioned above, step B, decomposes the preprocessed signal through a multi-channel filter bank and outputs a multi-channel signal, wherein the preprocessed signal is evenly divided into N channels with overlapping frequencies, and then the signal in each channel is sampled D times, and the sampling rate D≤N / 2, and the output multi-channel signal is Ch (1) , Ch (2) ……Ch (N) .

[0014] The aforementioned low-complexity and high-resolution howling detection and suppression method for digital hearing aids, step C, detects whether there is howling in the current ambient sound signal according to the energy distribution of the multi-channel signal, and obtains the howling detection result. The specific steps are as follows:

[0015] Step C1, calculate the energy E of each channel (1) 、E (2) ……E (N) And obtain the channel number m1 and energy value E of the maximum energy channel (m1) , and then obtain the channel number m2 and energy value E of the second largest energy channel (m2) ;

[0016] Step C2, setting the upper boundary h of the howling discrimination frequency band according to the hearing aid prior howling frequency band i and the lower boundary l o , and then determine whether m1 and m2 satisfy l o <m1<h i or o <m2<h i If not, it is determined that there is no howling. If satisfied, proceed to the subsequent steps;

[0017] Step C3: preset the weight W and compare the weight W with the maximum channel energy E (m1) , the energy of the second largest channel E (m2) And short-time energy I are weighted calculated, as shown in formula (1):

[0018] y0 = W (0) *I + W (1) *E (m1) +W (2) *E (m2) +b0(1)

[0019] Among them, y0 is the first operation result, W (0) , W (1) and W (2) are weights, b0 is the offset, if the first operation result y0>0, then record the maximum energy channel number m1 and the maximum energy channel signal sample point Ch (m1) And continue the subsequent steps, if the first operation result y0≤0, it is determined that there is no howling;

[0020] Step C4: record the maximum energy channel signal sample point Ch (m1) Perform a short-time Fourier transform, then calculate the maximum energy frequency point energy P and the weight W to obtain the second calculation result y1, as shown in formula (2),

[0021] y1=W (3) *P+W (4) *I+W (5) *E (m1) +W (6) *E (m2) +b1 (2)

[0022] Among them, W (3) , W (4) , W (5) and W (6) are all weights, b1 is an offset, if the second operation result y1>0, it is determined that howling is detected, if the second operation result y1≤0, it is determined that howling is not detected.

[0023] The aforementioned low-complexity and high-resolution howling detection and suppression method for digital hearing aids, step D, uses a host computer to determine whether the howling detection result is accurate. If the howling detection result is wrong, the weight adjustment is automatically performed. If the howling detection result is correct, the weight adjustment process is closed. The specific steps are as follows:

[0024] Step D1, if the first operation result y0>0, the host computer determines that howling does not occur. If the first operation result y0≤0, the host computer determines that howling exists and sets the parameter k0 to adjust the gradient and adjust the weight W. The specific steps are as follows, as shown in formula (3):

[0025] W (0) =W (0) - y0*k0*I,

[0026] W (1) =W (1) -y0*k0*E (m1) ,

[0027] W (2) =W (2) -y0*k0*E (m2) ,

[0028] b0´=b0-y0*k0(3)

[0029] Among them, W (0) ´, W (1) ´ and W (2) ´ are all adjusted weights, b0´ is the adjusted offset;

[0030] Step D2, if the second operation result y1>0, the host computer determines that howling has not occurred. If the second operation result y1≤0, the host computer determines that howling has occurred and sets k1 as the adjustment gradient to adjust the weight W. The specific steps are as follows, as shown in formula (4):

[0031] W (3) =W (3) -y1*k1*P,

[0032] W (4) =W (4) -y1*k1*I,

[0033] W (5) =W (5) -y1*k1*E (m1) ,

[0034] W (6) =W (6) -y1*k1*E (m2) ,

[0035] b1´=b1-y1*k1; (4)

[0036] Among them, W (3) ´, W (4) ´, W (5) ´ and W (6) ´ are all adjusted weights, and b1´ is the adjusted offset.

[0037] The low-complexity and high-resolution howling detection and suppression method for digital hearing aids mentioned above, step E, calculates the feedback gain based on the howling detection result and completes the howling suppression operation, wherein specifically the maximum energy channel sequence number m1 and the maximum energy channel signal sample point Ch obtained by detection are used (m1) To calculate the feedback gain, the specific steps are as follows:

[0038] Step E1, obtain the historical channel gain g of the hearing aid algorithm on channel m1 (m1) Processed channel signal Cg (m1) , where Cg (m1) is the channel m1 signal 4 milliseconds ago, g (m1) is the total gain of the algorithm corresponding to channel m1;

[0039] Step E2, calculate the feedback gain f(m1)´ of the historical frame m1 channel by the minimum mean square error method, as shown in formula (5),

[0040] f(m1)´=f(m1)+c*Cg (m1) *(Ch (m1) -f (m1) *Cg (m1) )(5)

[0041] Among them, f(m1) is the pre-measurement feedback gain, and c is the update step size;

[0042] Step E3: if y1>0, i.e. howling is detected, the channel gain is suppressed. The specific gain suppression process is: if the channel gain g (m1) >1 / abs(f (m1) ), then the channel gain g (m1) =1 / abs(f (m1) ).

[0043] A low-complexity, high-resolution howling detection and suppression system for a digital hearing aid comprises a signal acquisition module, a preprocessing module, a howling detection module, a detection and discrimination module and a howling suppression module, wherein the signal acquisition module is used to acquire sound signals from the surrounding environment and convert them into digital signals, and then preprocess the digital signals to obtain preprocessed signals; the preprocessing module is used to decompose the preprocessed signals through a multi-channel filter group and output a multi-channel signal; the howling detection module is used to detect whether there is howling in the current environmental sound signal according to the energy distribution of the multi-channel signal and obtain a howling detection result; the detection and discrimination module is used to use a host computer to determine whether the howling detection result is accurate; and the howling suppression module is used to calculate feedback gain based on the howling detection result and complete the howling suppression operation.

[0044] In the aforementioned low-complexity and high-resolution howling detection and suppression system for digital hearing aids, if the howling detection result in the detection and discrimination module is erroneous, automatic weight adjustment is performed; if the howling detection result is correct, the weight adjustment process is closed.

[0045] The beneficial effects of the present invention are as follows: a low-complexity, high-resolution howling detection and suppression method and system for a digital hearing aid of the present invention first collects sound signals from the surrounding environment and converts them into digital signals, then preprocesses the digital signals and obtains preprocessed signals, then decomposes the preprocessed signals through a multi-channel filter group and outputs a multi-channel signal, then detects whether the current environmental sound signal has howling according to the energy distribution of the multi-channel signal and obtains a howling detection result, then uses a host computer to determine whether the howling detection result is accurate, if the howling detection result is wrong, then automatically adjusts the weight, if the howling detection result is correct, then closes the weight adjustment process, then calculates the feedback gain based on the howling detection result and completes the howling suppression operation; the present invention realizes the function of reducing the computational complexity and improving the detection accuracy through multi-level processing and step-by-step refinement, and can be deployed and operated on a low-power and low-computing power platform with extremely small resource overhead through the multi-channel decomposition module and the filtering module in the digital hearing aid, and the preset weight adjustment through the weight adjustment mechanism ensures that the system stably and accurately detects and suppresses howling in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flow chart of a low-complexity and high-resolution howling detection and suppression method for a digital hearing aid of the present invention;

[0047] Figure 2 It is a schematic diagram of the signal howling detection and suppression principle of the digital hearing aid of the present invention;

[0048] Figure 3 is a flow chart of howling detection and weight adjustment of the present invention;

[0049] Figure 4 It is a howling suppression flow chart of the present invention. DETAILED DESCRIPTION

[0050] The present invention will be further described below in conjunction with the accompanying drawings.

[0051] like Figure 1 As shown, a low-complexity and high-resolution howling detection and suppression method for a digital hearing aid of the present invention comprises the following steps:

[0052] Step A, collects the sound signal of the surrounding environment and converts it into a digital signal, and then preprocesses the digital signal to obtain the preprocessed signal, wherein the preprocessing of the digital signal and obtaining the preprocessed signal specifically counts the short-time energy I of the broadband digital signal after filtering, and if the short-time energy I is greater than the set energy threshold Th, the signal gain is reduced during the collection and filtering process.

[0053] like Figure 2 As shown, step B, decomposes the preprocessed signal through a multi-channel filter bank and outputs a multi-channel signal, wherein the preprocessed signal is evenly divided into N channels with overlapping frequencies, and then the signal in each channel is sampled D times, and the sampling rate D≤N / 2, and the output multi-channel signal is Ch (1) , Ch (2) ……Ch (N) .

[0054] like Figure 3 As shown, step C, according to the energy distribution of the multi-channel signal, detects whether there is howling in the current environmental sound signal, and obtains the howling detection result. The specific steps are as follows:

[0055] Step C1, calculate the energy E of each channel (1) 、E (2) ……E (N) And obtain the channel number m1 and energy value E of the maximum energy channel (m1) , and then obtain the channel number m2 and energy value E of the second largest energy channel (m2) ;

[0056] Step C2, setting the upper boundary h of the howling discrimination frequency band according to the hearing aid prior howling frequency band i and the lower boundary l o , and then determine whether m1 and m2 satisfy l o <m1<h i or o <m2<h i If not, it is determined that there is no howling. If satisfied, proceed to the subsequent steps;

[0057] Step C3: preset the weight W and compare the weight W with the maximum channel energy E (m1) , the energy of the second largest channel E (m2) And short-time energy I are weighted calculated, as shown in formula (1):

[0058] y0 = W (0) *I + W (1) *E (m1) +W (2) *E (m2) +b0(1)

[0059] Among them, y0 is the first operation result, W (0) , W (1) and W (2) are weights, b0 is the offset, if the first operation result y0>0, then record the maximum energy channel number m1 and the maximum energy channel signal sample point Ch (m1) And continue the subsequent steps, if the first operation result y0≤0, it is determined that there is no howling;

[0060] Step C4: record the maximum energy channel signal sample point Ch (m1) Perform a short-time Fourier transform, then calculate the maximum energy frequency point energy P and the weight W to obtain the second calculation result y1, as shown in formula (2),

[0061] y1=W (3) *P+W (4) *I+W (5) *E (m1) +W (6) *E (m2) +b1 (2)

[0062] Among them, W (3) , W (4) , W (5) and W (6) are all weights, b1 is an offset, if the second operation result y1>0, it is determined that howling is detected, if the second operation result y1≤0, it is determined that howling is not detected.

[0063] like Figure 4 As shown, step D uses the host computer to determine whether the howling detection result is accurate. If the howling detection result is wrong, the weight adjustment is performed automatically. If the howling detection result is correct, the weight adjustment process is closed. The specific steps are as follows:

[0064] Step D1, if the first operation result y0>0, the host computer determines that howling does not occur. If the first operation result y0≤0, the host computer determines that howling exists and sets the parameter k0 to adjust the gradient and adjust the weight W. The specific steps are as follows, as shown in formula (3):

[0065] W (0) =W (0) - y0*k0*I,

[0066] W (1) =W (1) -y0*k0*E (m1) ,

[0067] W (2) =W (2) -y0*k0*E (m2) ,

[0068] b0´=b0-y0*k0(3)

[0069] Among them, W (0) ´, W (1) ´ and W (2) ´ are all adjusted weights, b0´ is the adjusted offset;

[0070] Step D2, if the second operation result y1>0, the host computer determines that howling has not occurred. If the second operation result y1≤0, the host computer determines that howling has occurred and sets k1 as the adjustment gradient to adjust the weight W. The specific steps are as follows, as shown in formula (4):

[0071] W (3) =W (3) -y1*k1*P,

[0072] W (4) =W (4) -y1*k1*I,

[0073] W (5) =W (5) -y1*k1*E (m1) ,

[0074] W (6) =W (6) -y1*k1*E (m2) ,

[0075] b1´=b1-y1*k1; (4)

[0076] Among them, W (3) ´, W (4) ´, W (5) ´ and W (6) ´ are all adjusted weights, and b1´ is the adjusted offset.

[0077] Step E, based on the howling detection result, the feedback gain is calculated and the howling suppression operation is completed, wherein specifically, the maximum energy channel sequence number m1 and the maximum energy channel signal sample point Ch obtained by detection are used. (m1)To calculate the feedback gain, the specific steps are as follows:

[0078] Step E1, obtain the historical channel gain g of the hearing aid algorithm on channel m1 (m1) Processed channel signal Cg (m1) , where Cg (m1) is the channel m1 signal 4 milliseconds ago, g (m1) is the total gain of the algorithm corresponding to channel m1;

[0079] Step E2, calculate the feedback gain f(m1)´ of the historical frame m1 channel by the minimum mean square error method, as shown in formula (5),

[0080] f(m1)´=f(m1)+c*Cg (m1) *(Ch (m1) -f (m1) *Cg (m1) )(5)

[0081] Among them, f(m1) is the pre-measurement feedback gain, and c is the update step size;

[0082] Step E3: if y1>0, i.e. howling is detected, the channel gain is suppressed. The specific gain suppression process is: if the channel gain g (m1) >1 / abs(f (m1) ), then the channel gain g (m1) =1 / abs(f (m1) ).

[0083] A low-complexity, high-resolution howling detection and suppression system for a digital hearing aid comprises a signal acquisition module, a preprocessing module, a howling detection module, a detection and discrimination module and a howling suppression module, wherein the signal acquisition module is used to acquire sound signals from the surrounding environment and convert them into digital signals, and then preprocess the digital signals to obtain preprocessed signals; the preprocessing module is used to decompose the preprocessed signals through a multi-channel filter group and output a multi-channel signal; the howling detection module is used to detect whether there is howling in the current environmental sound signal according to the energy distribution of the multi-channel signal and obtain a howling detection result; the detection and discrimination module is used to use a host computer to determine whether the howling detection result is accurate; and the howling suppression module is used to calculate feedback gain based on the howling detection result and complete the howling suppression operation.

[0084] Specifically, if the howling detection result in the detection and discrimination module is wrong, the weight adjustment is automatically performed, and if the howling detection result is correct, the weight adjustment process is closed.

[0085] In summary, a low-complexity, high-resolution howling detection and suppression method and system for a digital hearing aid of the present invention first collects sound signals from the surrounding environment and converts them into digital signals, then preprocesses the digital signals and obtains preprocessed signals, then decomposes the preprocessed signals through a multi-channel filter group and outputs a multi-channel signal, and then detects whether the current environmental sound signal has howling according to the energy distribution of the multi-channel signal and obtains a howling detection result, and then uses a host computer to determine whether the howling detection result is accurate. If the howling detection result is wrong, the weight adjustment is performed automatically. If the howling detection result is correct, the weight adjustment process is closed, and then the feedback gain is calculated based on the howling detection result and the howling suppression operation is completed; the present invention realizes the function of reducing the computational complexity and improving the detection accuracy through multi-level processing and gradual refinement, and the multi-channel decomposition module and the filtering module in the digital hearing aid can be deployed and run on a low-power and low-computing power platform with extremely small resource overhead, and the preset weight adjustment through the weight adjustment mechanism ensures that the system can stably and accurately detect and suppress howling in practical applications.

[0086] 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. A low-complexity and high-resolution howling detection and suppression method for digital hearing aids, characterized in that: The following steps are included: Step A, collecting sound signals from the surrounding environment and converting them into digital signals, and then preprocessing the digital signals to obtain preprocessed signals, wherein the preprocessing of the digital signals and obtaining the preprocessed signals specifically includes counting the short-time energy I of the broadband digital signals after filtering, and if the short-time energy I is greater than the set energy threshold Th, reducing the signal gain during the collection and filtering process; Step B, decompose the preprocessed signal through a multi-channel filter bank and output a multi-channel signal, wherein the preprocessed signal is evenly divided into N channels with overlapping frequencies, and then the signal in each channel is sampled D times, and the sampling rate D≤N / 2, and the output multi-channel signal is Ch (1) , Ch (2) ……Ch (N) ; Step C, detecting whether there is howling in the current ambient sound signal according to the energy distribution of the multi-channel signal, and obtaining a howling detection result, the specific steps are as follows: Step C1, calculate the energy E of each channel (1) 、E (2) ……E (N) And obtain the channel number m1 and energy value E of the maximum energy channel (m1) , and then obtain the channel number m2 and energy value E of the second largest energy channel (m2) ; Step C2, setting the upper boundary h of the howling discrimination frequency band according to the hearing aid prior howling frequency band i and the lower boundary l o , and then determine whether m1 and m2 satisfy l o <m1<h i or o <m2<h i If not, it is determined that there is no howling. If satisfied, proceed to the subsequent steps; Step C3: preset the weight W and compare the weight W with the maximum channel energy E (m1) , the energy of the second largest channel E (m2) And short-time energy I are weighted calculated, as shown in formula (1), y0=W (0) *I+W (1) *E (m1) +W (2) *E (m2) +b0(1) Among them, y0 is the first operation result, W (0) , W (1) and W (2) are weights, b0 is the offset, if the first operation result y0>0, then record the maximum energy channel number m1 and the maximum energy channel signal sample point Ch (m1) And continue the subsequent steps, if the first operation result y0≤0, it is determined that there is no howling; Step C4: record the maximum energy channel signal sample point Ch (m1) Perform a short-time Fourier transform, then calculate the maximum energy frequency point energy P and the weight W to obtain the second calculation result y1, as shown in formula (2), y1=W (3) *P+W (4) *I+W (5) *E (m1) +W (6) *E (m2) +b1(2) Among them, W (3) , W (4) , W (5) and W (6) are all weights, b1 is an offset, if the second operation result y1>0, it is determined that howling is detected, if the second operation result y1≤0, it is determined that howling is not detected; Step D, using the host computer to determine whether the howling detection result is accurate, if the howling detection result is wrong, then automatically adjust the weight, if the howling detection result is correct, then close the weight adjustment process; Step E: Calculate feedback gain based on the howling detection result and complete the howling suppression operation.

2. The low-complexity and high-resolution howling detection and suppression method for a digital hearing aid according to claim 1, characterized in that: Step D, using the host computer to determine whether the howling detection result is accurate, if the howling detection result is wrong, then automatically adjust the weights, if the howling detection result is correct, then close the weight adjustment process, the specific steps are as follows, Step D1, if the first operation result y0>0, the host computer determines that howling does not occur. If the first operation result y0≤0, the host computer determines that howling exists and sets the parameter k0 to adjust the gradient and adjust the weight W. The specific steps are as follows, as shown in formula (3): IN (0) ′=W (0) -y0*k0*I, IN (1) ′=W (1) -y0*k0*E (m1) , IN (2) ′=W (2) -y0*k0*E (m2) , b0′=b0-y0*k0(3) Among them, W (0) ′、W (1) ′ and W (2) ′ are all adjusted weights, b0′ is the adjusted offset; Step D2, if the second operation result y1>0, the host computer determines that howling has not occurred. If the second operation result y1≤0, the host computer determines that howling has occurred and sets k1 as the adjustment gradient to adjust the weight W. The specific steps are as follows, as shown in formula (4): IN (3) ′=W (3) -y1*k1*P, IN (4) ′=W (4) -y1*k1*I, IN (5) ′=W (5) -y1*k1*E (m1) , IN (6) ′=W (6) -y1*k1*E (m2) , b1′=b1-y1*k1;(4) Among them, W (3) ′、W (4) ′、W (5) ′ and W (6) ′ are all adjusted weights, and b1′ is the adjusted offset.

3. The low-complexity and high-resolution howling detection and suppression method for a digital hearing aid according to claim 2, characterized in that: Step E, based on the howling detection result, the feedback gain is calculated and the howling suppression operation is completed, wherein specifically, the maximum energy channel sequence number m1 and the maximum energy channel signal sample point Ch obtained by detection are used. (m1) To calculate the feedback gain, the specific steps are as follows: Step E1, obtain the historical channel gain g of the hearing aid algorithm on channel m1 (m1) Processed channel signal Cg (m1) , where Cg (m1) is the channel m1 signal 4 milliseconds ago, g (m1) is the total gain of the algorithm corresponding to channel m1; Step E2, calculate the feedback gain f(m1)′ of the historical frame m1 channel by the minimum mean square error method, as shown in formula (5), f(m1)′=f(m1)+c*Cg (m1) *(Ch (m1) -f (m1) *Cg (m1) )(5) Among them, f(m1) is the pre-measurement feedback gain, and c is the update step size; Step E3: if y1>0, i.e. howling is detected, the channel gain is suppressed. The specific gain suppression process is: if the channel gain g (m1) >1 / abs(f (m1) ), then the channel gain g (m1) =1 / abs(f (m1) ).

4. A low-complexity and high-resolution howling detection and suppression system for a digital hearing aid, wherein the howling detection and suppression process of the howling detection and suppression system is based on the howling detection and suppression method according to any one of claims 1 to 3, characterized in that: It includes a signal acquisition module, a preprocessing module, a howling detection module, a detection and discrimination module and a howling suppression module. The signal acquisition module is used to collect sound signals from the surrounding environment and convert them into digital signals, and then preprocess the digital signals to obtain preprocessed signals; The preprocessing module is used to decompose the preprocessed signal through a multi-channel filter bank and output a multi-channel signal; The howling detection module is used to detect whether there is howling in the current environmental sound signal according to the energy distribution of the multi-channel signal and obtain a howling detection result; The detection and discrimination module is used to use the host computer to determine whether the howling detection result is accurate; The howling suppression module is used to calculate the feedback gain based on the howling detection result and complete the howling suppression operation.

5. The low-complexity and high-resolution howling detection and suppression system for a digital hearing aid according to claim 4, characterized in that: If the howling detection result in the detection and discrimination module is wrong, the weight adjustment is automatically performed; if the howling detection result is correct, the weight adjustment process is closed.

Images