A digital hearing aid howl suppression method and system

Abstract

The application relates to the technical field of hearing aids, in particular to a digital hearing aid howling suppression method and system, which comprises the following steps: determining an energy distribution characteristic value; determining the noise possibility of each to-be-processed signal based on the distribution of all high-energy values and derived values of each to-be-processed signal and a preset number of to-be-processed signals before the to-be-processed signal in a frequency domain and the rate at which each peak value in each to-be-processed signal drops to the adjacent next valley value, and determining an over-reduction index in combination with the energy distribution characteristic value; and suppressing howling noise in a speech signal by using a spectral subtraction method based on the over-reduction index. The application dynamically analyzes the frequency domain energy distribution and the time domain oscillation characteristics, adaptively adjusts the over-reduction index of the spectral subtraction method, solves the problem of poor adaptability of a traditional howling suppression method, and improves the howling suppression effect of a digital hearing aid in a complex acoustic environment.

Description

Technical Field

[0001] This application relates to the field of hearing aid technology, specifically to a method and system for suppressing feedback in digital hearing aids. Background Technology

[0002] Digital hearing aids convert received speech signals from analog to digital. Through digital signal processing algorithms integrated into a digital signal processing chip, the speech signal is refined to improve its signal-to-noise ratio, thus assisting hearing. However, the speech signal output from the digital hearing aid's speaker is prone to leakage through gaps between the earpiece and the ear canal, or through internal vents, and can be picked up again by the microphone, creating acoustic feedback. This causes speech signal distortion, and in severe cases, complete distortion. When certain gain and phase conditions are met, the output speech signal can produce howling, affecting the performance of the digital hearing aid.

[0003] Currently, feedback suppression in digital hearing aids commonly employs a combination of notch filtering and spectral subtraction. Notch filtering suppresses feedback by attenuating the gain at the feedback frequency, while spectral subtraction aims to reduce the resulting speech distortion. However, this approach has significant drawbacks: the time-varying nature of speech signals leads to a random distribution of feedback frequencies, making it difficult for notch filtering to respond comprehensively; simultaneously, notch filtering itself introduces new time-varying noise, and existing spectral subtraction methods rely on fixed over-attenuation coefficients, failing to dynamically adjust according to the randomness and time-varying nature of feedback. This results in unsatisfactory suppression effects across different frequency bands, reducing the effectiveness of digital hearing aids in suppressing feedback noise. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a method and system for suppressing feedback in digital hearing aids, the specific technical solution of which is as follows:

[0005] In a first aspect, embodiments of this application provide a method for suppressing feedback in a digital hearing aid, the method comprising the following steps:

[0006] Acquire speech signals from a digital hearing aid and segment the speech signals into multiple signals to be processed;

[0007] Based on the energy distribution of each signal to be processed in the frequency domain, the energy value of each signal to be processed in the frequency domain is divided into high and low energy values; based on the difference in the distribution between high and low energy values ​​of each signal to be processed in the frequency domain, and the degree of disorder of all energy values, the energy distribution characteristic value of each signal to be processed is determined.

[0008] determine a forward evaluation index of each to-be-processed signal based on the distribution of all high-energy values and derived values of each to-be-processed signal and a preset number of to-be-processed signals before the to-be-processed signal in the frequency domain; determine an oscillation evaluation parameter of each to-be-processed signal based on a rate at which each peak value in each to-be-processed signal drops to an adjacent next valley value; determine a backward evaluation index of each to-be-processed signal based on a difference between the oscillation evaluation parameters of each to-be-processed signal and a preset number of to-be-processed signals after the to-be-processed signal, and determine a noise-likelihood of each to-be-processed signal in combination with the forward evaluation index;

[0009] determine an over-reduction index of each to-be-processed signal based on the energy distribution characteristic value and the noise-likelihood, and suppress the howling noise in the speech signal by using a spectral subtraction method based on the over-reduction index.

[0010] Preferably, the dividing of the energy values of each to-be-processed signal in the frequency domain into high-energy values and low-energy values comprises:

[0011] calculate a mean value of all energy values of each to-be-processed signal in the frequency domain, denoted as a global threshold value, and regard all energy values greater than the global threshold value in each to-be-processed signal as high-energy values, and regard the remaining energy values as low-energy values.

[0012] Preferably, the method for determining the energy distribution characteristic value of each to-be-processed signal comprises:

[0013] calculate a proportion of the number of high-energy values with a frequency greater than a fundamental frequency in the to-be-processed signal in the total number of high-energy values of the to-be-processed signal, denoted as a high-energy distribution value; and calculate a proportion of the number of low-energy values with a frequency greater than the fundamental frequency in the to-be-processed signal in the total number of low-energy values of the to-be-processed signal, denoted as a low-energy distribution value.

[0014] calculate a difference value between the high-energy distribution value and the low-energy distribution value, and regard a result of positively fusing the information entropy of all energy values of each to-be-processed signal with the difference value as the energy distribution characteristic value of each to-be-processed signal.

[0015] Preferably, the method for determining the forward evaluation index of each to-be-processed signal comprises:

[0016] respectively acquire a mean value, a cumulative sum, a maximum value and a number of all high-energy values of each to-be-processed signal in the frequency domain;

[0017] respectively fit the high-energy value mean, the high-energy value cumulative sum, the high-energy value maximum value and the high-energy value number of each to-be-processed signal and a preset number of to-be-processed signals before the to-be-processed signal, to obtain four fitting straight lines, and respectively calculate a sum value of slopes of the four fitting straight lines and a sum value of fitting errors.

[0018] The forward evaluation index of each to-be-processed signal is negatively correlated with the slope and value of the fitting straight line corresponding to the to-be-processed signal and the fitting error and value of the fitting straight line.

[0019] Preferably, the method for determining the oscillation evaluation parameter of each to-be-processed signal comprises the following steps:

[0020] The rate at which each peak value in each to-be-processed signal drops to the adjacent next valley value is recorded as the drop rate of the peak value, and the average of the drop rate difference between all peak values and the adjacent next peak values is taken as the oscillation evaluation parameter of each to-be-processed signal.

[0021] Preferably, the backward evaluation index of each to-be-processed signal is the sum of the oscillation evaluation parameter difference between each to-be-processed signal and the subsequent preset number of to-be-processed signals.

[0022] Preferably, the noise possibility of each to-be-processed signal is the result of the positive fusion of the normalized value of the forward evaluation index and the normalized value of the backward evaluation index of each to-be-processed signal.

[0023] Preferably, the expression of the over-reduction index of each to-be-processed signal is: ; in the formula, , respectively represent the over-reduction index of the i th and the i-1 th to-be-processed signal, wherein the initial value of the over-reduction index is a preset value; , respectively represent the spectral subtraction smoothing coefficient of the i th and the i-1 th to-be-processed signal, wherein the spectral subtraction smoothing coefficient of each to-be-processed signal is the normalized value of the energy distribution characteristic value and the normalized average of the noise possibility; represents a preset adjustment step.

[0024] Preferably, the method for suppressing the howling noise in the speech signal by using the spectral subtraction method comprises:

[0025] Obtaining the estimated noise spectrum of each frame of to-be-processed signal, taking the spectrum of all to-be-processed signals and the estimated noise spectrum thereof as the input of the spectral subtraction method, taking the over-reduction index of all to-be-processed signals as the over-checking factor in the spectral subtraction method, and outputting all to-be-processed signals after noise reduction. In the second aspect, the embodiments of the present application also provide a digital hearing aid howling suppression system, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the digital hearing aid howling suppression method of any one of the above aspects when executing the computer program.

[0026] The present application has at least the following beneficial effects:

[0027] The application analyzes the energy distribution characteristics of the to-be-processed signal in the frequency domain, innovatively combines the information entropy and the difference between the high and low energy distribution values, constructs an energy distribution characteristic value, accurately identifies the typical features of the howling residual noise, effectively distinguishes the noise and the effective speech, and helps to improve the howling suppression effect of the digital hearing aid in a complex acoustic environment; further, the application fuses the time domain oscillation characteristics and the frequency domain energy attenuation characteristics, constructs a forward and backward bidirectional evaluation index, realizes accurate identification and quantitative evaluation of the howling residual noise, and helps to improve the howling suppression effect of the digital hearing aid; finally, the energy distribution characteristic value and the noise probability are fused to construct a spectral subtraction smoothing coefficient, and the subtraction index is adaptively adjusted based on the dynamic change of the coefficient between adjacent frames, the real-time optimization of the spectral subtraction noise suppression strength is realized, the original quality of the speech signal is maximized while effectively suppressing the howling residual noise, and the howling suppression effect of the digital hearing aid in a complex acoustic environment is improved. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0029] Figure 1 A step flow chart of a digital hearing aid howling suppression method provided by an embodiment of the present application;

[0030] Figure 2 A spectral subtraction smoothing coefficient extraction process schematic diagram provided by an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation, structure, features and effects of the digital hearing aid howling suppression method and system according to the present application are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0033] The specific scheme of the digital hearing aid howling suppression method and system provided by the present application is specifically described below in combination with the drawings.

[0034] Referring to Figure 1 which shows a flow chart of steps of a digital hearing aid howling suppression method provided by an embodiment of the present application, the method comprising the following steps:

[0035] Step S1: obtaining a speech signal in a digital hearing aid.

[0036] In the embodiment, the sampling frequency of the speech signal in the digital hearing aid is 44.1 kHz. For the convenience of data processing, the speech signal is processed by frame, and the length of each speech frame is set to 25 ms. Meanwhile, in order to reduce distortion in the speech signal processing process, the embodiment adopts an alternate frame method, that is, there is an overlap between adjacent two speech frames, and the frame shift is set to 1 / 2 of the length of the speech frame.

[0037] In order to reduce the spectral leakage and boundary effect caused by the frame processing of the speech signal, the embodiment adopts a window function to perform windowing processing on each speech frame.

[0038] Further, since the howling is caused by the amplified sound of the hearing aid being transmitted to the microphone through space or structure vibration, and then becoming an input signal to be circularly amplified to cause audio resonance, therefore, the embodiment first uses a notch method to perform acoustic feedback suppression on each speech frame. Specifically, the embodiment uses an adaptive notch filter (ANF) to perform acoustic feedback suppression on each speech frame, and each speech frame after acoustic feedback suppression is recorded as each to-be-processed signal.

[0039] Step S2: based on the energy distribution of each to-be-processed signal in the frequency domain, dividing the energy value of each to-be-processed signal in the frequency domain into high and low energy values; based on the difference between the high energy value and the low energy value of each to-be-processed signal in the frequency domain, and the degree of confusion of all energy values, determining the energy distribution characteristic value of each to-be-processed signal.

[0040] The cause of howling lies in the acoustic feedback loop: the sound output by the hearing aid loudspeaker leaks to the microphone through the air or structure vibration path, forming a closed loop, when the gain and phase of the loop meet the oscillation condition, the signal is circularly amplified, and finally a self-excited oscillation is triggered at a specific frequency, that is, howling. Howling can be divided into internal howling and external howling according to its cause. In the embodiment, the spectrum subtraction method is used to suppress howling, and the effectiveness of the spectrum subtraction method is based on the following specific premise: the target noise is irrelevant to the speech signal, and its spectral characteristics can be statistically or estimated.

[0041] Since the to-be-processed signals obtained in step S1 are superpositions of effective speech signals and residual noise introduced by the acoustic feedback suppression, compared with the effective speech signals, the residual noise shows stronger randomness and time variability in the frequency domain, and the energy distribution of the residual noise in different speech frames and different frequency points is significantly different. Based on this characteristic, the energy values of each to-be-processed signal in the frequency domain are divided into high and low energy values based on the energy distribution of each to-be-processed signal in the frequency domain. Based on the difference between the high energy values and the low energy values of each to-be-processed signal in the frequency domain and the degree of confusion of all energy values, the energy distribution characteristic value of each to-be-processed signal is determined to accurately identify and filter the frequency domain region contaminated by noise. Specifically:

[0042] First, the energy values of each to-be-processed signal in the frequency domain are divided into high and low energy values based on the energy distribution of each to-be-processed signal in the frequency domain. Specifically:

[0043] In this embodiment, the mean of all energy values of each to-be-processed signal in the frequency domain is calculated, denoted as a global threshold. All energy values greater than the global threshold in each to-be-processed signal are regarded as high energy values, and the remaining energy values are regarded as low energy values.

[0044] It is further explained that the energy values of each to-be-processed signal in the frequency domain can be obtained by using a short-time Fourier transform algorithm to obtain a time-frequency graph of each to-be-processed signal. The horizontal coordinate of the time-frequency graph is time, and the vertical coordinate is frequency. The energy size of each frequency point is represented by the color depth in the two-dimensional time-frequency graph. Generally, the deeper the color, the higher the energy. Therefore, the size of the color value can be used to represent the size of the energy value. In this embodiment, taking the color from black to white as an example, black (pixel value 0) represents the minimum value of the energy value, and white (pixel value 255) represents the maximum value of the energy value. It should be noted that in other embodiments, different colors can be used to represent different sizes of energy values, and this embodiment does not make special limitations.

[0045] The short-time Fourier transform algorithm is a known technology, and the specific process of obtaining the time-frequency graph of the signal using the short-time Fourier transform algorithm is not repeated here.

[0046] Further, the energy distribution characteristic value of each to-be-processed signal is determined based on the difference between the high energy values and the low energy values of each to-be-processed signal in the frequency domain and the degree of confusion of all energy values, which is used to represent the characteristics of the energy value distribution in each to-be-processed signal. Specifically:

[0047] In the embodiment, a ratio of a number of high energy values of each to-be-processed signal at all frequencies greater than the fundamental frequency in the frequency domain in a total number of all high energy values of the to-be-processed signal is calculated, and is recorded as a high energy distribution value; a ratio of a number of low energy values of each to-be-processed signal at all frequencies greater than the fundamental frequency in the frequency domain in a total number of all low energy values of the to-be-processed signal is calculated, and is recorded as a low energy distribution value.

[0048] A difference value of the high energy distribution value and the low energy distribution value is calculated, and a result of forward fusion of information entropy of all energy of each to-be-processed signal in the frequency domain and the difference value is taken as an energy distribution feature value of each to-be-processed signal.

[0049] The fundamental frequency is a known technology, and a specific acquisition method thereof will not be described herein.

[0050] It should be understood that the forward fusion refers to combining two or more indexes together through addition or multiplication or the like, so as to obtain a comprehensive index, thereby more comprehensively and accurately evaluating a phenomenon or a problem. The fusion method is not limited to simple arithmetic operation, but can also include more complex statistical models and analysis methods, and the implementer can select them according to specific conditions, and the embodiment does not have special limitations.

[0051] Preferably, as an implementation manner, in the embodiment, a product of the information entropy of all energy of each to-be-processed signal in the frequency domain and the difference value is taken as the result of forward fusion of the information entropy of all energy of each to-be-processed signal in the frequency domain and the difference value. In actual application, as other implementation manners, the implementer can also use other forward fusion methods such as addition or weighted fusion according to specific conditions, and the embodiment does not have special limitations.

[0052] According to the energy distribution feature value of each to-be-processed signal, it can be understood that the energy distribution feature value is used to represent the noise distribution characteristic of the to-be-processed signal in the frequency domain. The information entropy of all energy of the to-be-processed signal in the frequency domain reflects the randomness of the energy distribution, and the difference value of the high energy distribution value and the low energy distribution value is a bias factor of the high frequency domain energy distribution, and reflects the significance of the high frequency noise. If the information entropy of all energy of the current to-be-processed signal in the frequency domain is greater, it indicates that the noise content in the current to-be-processed signal is high. At the same time, if the difference value of the high energy distribution value and the low energy distribution value of the current to-be-processed signal is smaller, it indicates that the proportion of the high frequency energy in the current to-be-processed signal is higher, which is a typical feature value of the howling residual noise, and indicates that the possibility of the existence of the howling residual noise in the current to-be-processed signal is greater, and therefore, the corresponding energy distribution noise is greater.

[0053] Conversely, if the information entropy of all energy of the current signal to be processed in the frequency domain is smaller, it indicates that the energy distribution in the current signal to be processed is more concentrated, the randomness is lower, the signal is closer to pure speech, and the noise content is lower. At the same time, if the difference between the high energy distribution value and the low energy distribution value of the current signal to be processed is larger, it indicates that the energy distribution in the current signal to be processed in the frequency domain is more dispersed, and the low frequency energy occupies the dominant position, which does not conform to the characteristics of the high frequency concentration of the howling residual noise, and indicates that the possibility of the current signal to be processed containing howling residual noise is smaller. Therefore, the corresponding energy distribution noise is smaller.

[0054] So far, by analyzing the energy distribution characteristics of the signal to be processed in the frequency domain, the energy distribution characteristic value is constructed by innovatively combining the information entropy and the difference between the high and low energy distribution values, so as to accurately identify the typical characteristics of the howling residual noise, thereby effectively distinguishing the noise and the effective speech, and helping to improve the howling suppression effect of the digital hearing aid in the complex acoustic environment.

[0055] Step S3: determining a forward evaluation index of each signal to be processed based on the distribution of all high energy values and derived values thereof of each signal to be processed and a preset number of signals to be processed before the signal to be processed in the frequency domain; determining an oscillation evaluation parameter of each signal to be processed based on the rate at which each peak value in each signal to be processed drops to its adjacent next valley value; determining a backward evaluation index of each signal to be processed based on the difference between the oscillation evaluation parameters of each signal to be processed and a preset number of signals to be processed after the signal to be processed, and determining the noise-containing possibility of each signal to be processed in combination with the forward evaluation index.

[0056] The essence of howling is the self-excited oscillation generated by the closed-loop system formed by the acoustic feedback path in the digital hearing aid at a specific frequency point. After the preliminary acoustic feedback suppression, the energy of the oscillation signal is significantly attenuated, but the residual time-varying noise still presents characteristics different from regular speech signals in the time-frequency domain.

[0057] Specifically, the howling is manifested as energy mutation and sharp resonance peak in the frequency domain, that is, at the moment when the howling is suppressed, the energy of the narrowband frequency point originally lifted due to self-oscillation will mutate, and by comparing and analyzing the spectrum of the historical signal to be processed, it can be observed that the energy values in the howling frequency and its adjacent frequency band will sharply decrease regardless of the frequency points, and at the same time, due to the physical effect of acoustic resonance, the spectrum corresponding to the howling will contain one or more sharp spectral peaks, which is a significant frequency domain marker distinguishing the howling signal from wideband noise and general speech signal; in addition, from the time domain perspective, the howling is manifested as damped oscillation and transient decay, that is, after the howling signal is suppressed, its energy is not instantaneously zero, but enters a damped oscillation process, which means that the amplitude envelope of the continuous multiple signals to be processed after the current signal to be processed will present an exponential decay oscillation centered on the resonance frequency of the howling, which is essentially different from the natural start and release process of the speech signal.

[0058] Therefore, based on the above analysis, the embodiment determines the forward evaluation index of each signal to be processed based on the distribution of all high energy values and derived values of each signal to be processed and the preset number of signals to be processed before it in the frequency domain; determines the oscillation evaluation parameter of each signal to be processed based on the rate at which each peak value in each signal to be processed decreases to its adjacent next valley value; determines the backward evaluation index of each signal to be processed based on the difference between the oscillation evaluation parameters of each signal to be processed and the preset number of signals to be processed after it, and determines the noise-containing probability of each signal to be processed in combination with the forward evaluation index to identify the howling residual noise, and the specific process is as follows:

[0059] In the embodiment, first, the forward evaluation index of each signal to be processed is determined based on the distribution of all high energy values and derived values of each signal to be processed and the preset number of signals to be processed before it in the frequency domain, specifically:

[0060] The embodiment calculates the mean and cumulative sum of all high energy values of each signal to be processed in the frequency domain, and counts the maximum value and the number of all high energy values of each signal to be processed in the frequency domain;

[0061] Further, the high energy value mean, the high energy value cumulative sum, the high energy value maximum, and the high energy value number of each signal to be processed and the preset number of signals to be processed before it are fitted respectively to obtain four fitting straight lines, and the sum of the slopes of the four fitting straight lines and the sum of the fitting errors of the four fitting straight lines are calculated respectively;

[0062] The forward evaluation index of each signal to be processed is negatively correlated with the sum of the slopes of the corresponding fitting straight lines and the sum of the fitting errors of the fitting straight lines.

[0063] It should be noted that the preset number is artificially set, and the preset number in the embodiment is 20. The preset number of signals to be processed is adjacent and continuous to each of the signals to be processed. In actual application, as other implementation manners, the implementer can set the value of the preset number according to the specific situation, and the embodiment does not have special limitations.

[0064] It should be noted that there are many common fitting algorithms. In the embodiment, the least square method is used to fit the high energy value mean, the high energy value accumulation, the high energy value maximum, and the high energy value number. In actual application, as other implementation manners, the implementer can use other fitting methods such as polynomial function fitting according to the specific situation, and the embodiment does not have special limitations.

[0065] The calculation method of the fitting error and the least square method are known technologies, and the calculation process of the fitting error and the specific process of fitting the data by using the least square method are not repeated.

[0066] It should be understood that the negative correlation indicates that the dependent variable decreases with the increase of the independent variable, and the dependent variable increases with the decrease of the independent variable, which can be a subtraction relationship or a division relationship, etc., which is determined by actual application.

[0067] Preferably, as an implementation manner, the expression of the forward evaluation index of each signal to be processed in the embodiment is: In the formula, represents the forward evaluation index of the i-th signal to be processed; represents the sum of the slopes of the four fitted straight lines corresponding to the i-th signal to be processed; represents the sum of the fitting errors of the four fitted straight lines corresponding to the i-th signal to be processed; exp( ) represents an exponential function with a natural constant as the base; represents a preset constant greater than 0, which is used to prevent the denominator from being 0, The value of is artificially set, and the value of in the embodiment is The value of is 0.01, and in actual application, as other implementation manners, the implementer can set it according to the specific situation on the premise of ensuring that the denominator is not 0 and does not excessively affect the calculation result, and the embodiment does not have special limitations.

[0068] According to the forward evaluation index of each to-be-processed signal, it can be understood that the forward evaluation index is used to characterize the noise energy in the to-be-processed signal and its historical to-be-processed signal in time sequence in a decay trend; the sum of the slopes of the four fitting straight lines corresponding to the to-be-processed signal reflects the decay rate of the noise energy; the sum of the fitting errors of the four fitting straight lines corresponding to the to-be-processed signal reflects the linear consistency of the energy fitting, that is, the stability and consistency of the energy decay process; if the sum of the slopes of the four fitting straight lines corresponding to the current to-be-processed signal is larger, it means that the energy of the current to-be-processed signal and the to-be-processed signal before it decays faster, and the corresponding forward evaluation index is smaller, which means that the noise of the current to-be-processed signal is rapidly decaying. At the same time, if the sum of the fitting errors of the four fitting straight lines corresponding to the current to-be-processed signal is smaller, it means that the energy decay process is very smooth and monotonous, and there is no fast and slow, which is another characteristic of the decay of the howling residual noise. This means that the possibility of the existence of howling in the current to-be-processed signal is larger, and the corresponding forward evaluation index is larger, which means that the noise in the current to-be-processed signal is more likely to be howling residual noise.

[0069] On the contrary, if the sum of the slopes of the four fitting straight lines corresponding to the current to-be-processed signal is smaller, it means that the energy of the current to-be-processed signal and the to-be-processed signal before it decays slower, or even tends to be stable or appears energy rebound, and the corresponding forward evaluation index is larger, which means that the energy of the current to-be-processed signal does not present a typical decay characteristic. At the same time, if the sum of the fitting errors of the four fitting straight lines corresponding to the current to-be-processed signal is larger, it means that the energy decay process is volatile and has poor linear consistency, and the energy changes in a chaotic manner, which does not conform to the characteristic of smooth decay of the howling residual noise. This means that the possibility of the existence of howling in the current to-be-processed signal is smaller, and the corresponding forward evaluation index is smaller, which means that the noise in the current to-be-processed signal is more likely to be random background noise rather than howling residual noise.

[0070] Further, the embodiment determines an oscillation evaluation parameter of each to-be-processed signal based on the rate at which each peak in each to-be-processed signal drops to its adjacent next valley.

[0071] In the embodiment, the rate at which each peak in each to-be-processed signal drops to its adjacent next valley is recorded as the drop rate of each peak, and the average of the difference between the drop rates of all peaks and their adjacent next peaks is taken as the oscillation evaluation parameter of each to-be-processed signal.

[0072] It is additionally explained that the calculation of the drop rate is the result of the difference between each peak and its adjacent next valley divided by the time interval therebetween.

[0073] According to the oscillation evaluation parameter of each to-be-processed signal, it can be understood that the oscillation evaluation parameter is used to represent the damped oscillation characteristics of the to-be-processed signal after the howling suppression in the time domain. The greater the difference between the falling rate of all peak values of the to-be-processed signal and the adjacent next peak value, the faster the oscillation amplitude decays with time, which corresponds to the exponential decay characteristics after the howling is suppressed. Therefore, the greater the corresponding oscillation evaluation parameter, the more significant the damped oscillation in the current to-be-processed signal, that is, the howling noise residue.

[0074] On the contrary, the smaller the difference between the falling rate of all peak values of the to-be-processed signal and the adjacent next peak value, the slower the oscillation amplitude decays with time, or even there is no obvious decay trend, and the amplitude remains stable or randomly fluctuates, which does not conform to the exponential decay characteristics after the howling is suppressed. Therefore, the smaller the corresponding oscillation evaluation parameter, the less significant the damped oscillation in the current to-be-processed signal, that is, the signal is more likely to be a stationary speech or random noise rather than a howling noise residue.

[0075] Further, the embodiment determines the backward evaluation index of each to-be-processed signal based on the difference between the oscillation evaluation parameters of each to-be-processed signal and the preset number of to-be-processed signals after the to-be-processed signal, and determines the noise possibility of each to-be-processed signal in combination with the forward evaluation index. Specifically,

[0076] In the embodiment, the sum of the difference between the oscillation evaluation parameters of each to-be-processed signal and the preset number of to-be-processed signals after the to-be-processed signal is used as the backward evaluation index of each to-be-processed signal.

[0077] It should be noted that the value of the preset number is artificially set, and the value of the preset number in the embodiment is 20. The preset number of to-be-processed signals is adjacent and continuous to the to-be-processed signal. In actual application, the implementer can set the value of the preset number according to the specific situation, and the embodiment does not have special limitations.

[0078] According to the backward evaluation index of each to-be-processed signal, it can be understood that the backward evaluation index is used to represent the decay consistency of the oscillation characteristics of the to-be-processed signal and the subsequent to-be-processed signal. The greater the difference between the oscillation evaluation parameters of the current to-be-processed signal and the preset number of to-be-processed signals after the to-be-processed signal, the more significant the oscillation intensity of the current to-be-processed signal is higher than that of the subsequent to-be-processed signal, which conforms to the physical law that the oscillation amplitude decays with time after the howling is suppressed. Therefore, the greater the corresponding backward evaluation index, the more the current to-be-processed signal is the starting point of the howling noise.

[0079] On the contrary, if the difference between the oscillation evaluation parameter of the current to-be-processed signal and the preset number of to-be-processed signals after the current to-be-processed signal is smaller, it indicates that the oscillation intensity of the current to-be-processed signal is similar to or has no obvious difference with the subsequent to-be-processed signal, and the oscillation phenomenon does not show a trend of attenuation over time, which does not conform to the physical law that the post-oscillation gradually weakens after the howling suppression, and thus the smaller the corresponding backward evaluation index is, the more likely the oscillation characteristic is caused by a stationary speech signal or a continuous random background noise.

[0080] Further, in the embodiment, the positive fusion result of the normalized value of the forward evaluation index and the normalized value of the backward evaluation index of each to-be-processed signal is taken as the noise-possibility degree of each to-be-processed signal.

[0081] Preferably, as an implementation manner, in the embodiment, the sum of the normalized value of the forward evaluation index and the normalized value of the backward evaluation index of each to-be-processed signal is taken as the noise-possibility degree of each to-be-processed signal, and in actual application, as other implementation manners, the implementer can also adopt other positive fusion methods such as multiplication according to specific conditions, and the embodiment does not have special limitation.

[0082] According to the noise-possibility degree of each to-be-processed signal, the noise probability reflects the possibility of containing the residual noise caused by the howling suppression in the to-be-processed signal, and is used to represent the noise pollution degree of the current to-be-processed signal; if the forward evaluation index of the current to-be-processed signal is larger, it indicates that the probability of containing noise in the current to-be-processed signal is larger, and thus the noise-possibility degree is larger; at the same time, if the backward evaluation index of the current to-be-processed signal is larger, it indicates that the current to-be-processed signal triggers strong time-domain oscillation, and the oscillation presents an obvious attenuation trend in the subsequent to-be-processed signal, which is the typical oscillation attenuation characteristic left in the time domain after the howling is produced and then suppressed, and this means that the possibility of containing the howling residual noise in the current to-be-processed signal is larger, and thus the noise-possibility degree is larger.

[0083] On the contrary, if the forward evaluation index of the current to-be-processed signal is smaller, it indicates that the probability of containing noise in the current to-be-processed signal is smaller, and thus the noise-possibility degree is smaller; at the same time, if the backward evaluation index of the current to-be-processed signal is smaller, it indicates that the current to-be-processed signal does not trigger strong time-domain oscillation, or the oscillation phenomenon does not present an obvious attenuation trend in the subsequent to-be-processed signal, which does not conform to the oscillation attenuation characteristic after the howling is suppressed, and this means that the possibility of containing the howling residual noise in the current to-be-processed signal is smaller, and thus the noise-possibility degree is smaller.

[0084] So far, the embodiment fuses the time domain oscillation characteristics and the frequency domain energy decay characteristics, constructs forward and backward bidirectional evaluation indexes, realizes accurate identification and quantitative evaluation of the howling residual noise, and helps to improve the suppression effect of the digital hearing aid howling.

[0085] Step S4: determining the over-reduction index of each to-be-processed signal based on the energy distribution characteristic value and the noise possibility degree; and suppressing the howling noise in the speech signal by using the spectral subtraction method based on the over-reduction index.

[0086] Based on the energy distribution characteristic value obtained in step S2 and the noise possibility degree obtained in step S3, the over-reduction index of each to-be-processed signal is determined, and further, the howling noise in the speech signal is suppressed by using the spectral subtraction method based on the over-reduction index. Specifically,

[0087] As an implementation mode, in the embodiment, the expression of the over-reduction index of the i-th to-be-processed signal is as follows: ; in the formula, wherein the over-reduction index of the (i-1)-th to-be-processed signal is represented by , and the initial value of the over-reduction index is a preset value; , wherein the spectral subtraction smoothing coefficients of the i-th and (i-1)-th to-be-processed signals are represented by and respectively, wherein the spectral subtraction smoothing coefficients of each to-be-processed signal are the normalized values of the energy distribution characteristic value and the normalized mean of the noise possibility degree; wherein the preset adjustment step length is represented by .

[0088] Preferably, the spectral subtraction smoothing coefficient extraction process provided in the embodiment is as shown in Figure 2 .

[0089] It should be noted that the value of the preset adjustment step length is artificially set, and the value range thereof is generally [0.5, 1]. In the embodiment, the value of the preset adjustment step length is 0.5. In actual application, as other implementation modes, the implementer can also set it by himself / herself according to the specific situation, and the embodiment does not make special limitation.

[0090] It should be further noted that the value of the preset value is also artificially set. In the embodiment, the value of the preset value is 1. In actual application, as other implementation modes, the implementer can also set it by himself / herself according to the specific situation, and the embodiment does not make special limitation.

[0091] Based on the over-subtraction exponent for each signal to be processed, it can be understood that the over-subtraction exponent is used to dynamically adjust the noise suppression strength of spectral subtraction. The spectral subtraction smoothing coefficient reflects the noise significance of the frequency domain energy distribution and time domain oscillation characteristics. If the difference between the spectral subtraction smoothing coefficient of the current signal to be processed and its adjacent previous signal to be processed is larger, it indicates that the noise content in the current signal to be processed is higher, the high-frequency energy is concentrated, and the oscillation attenuation is significant. Therefore, the corresponding over-subtraction exponent increases, enhancing the denoising strength of spectral subtraction. Conversely, if the difference between the spectral subtraction smoothing coefficient of the current signal to be processed and its adjacent previous signal to be processed is smaller, it indicates that the noise content in the current signal to be processed is lower, the frequency domain energy distribution is more stable, the time domain oscillation characteristics are not significant, and the signal is closer to clean speech. Therefore, the corresponding over-subtraction exponent decreases, reducing the denoising strength of spectral subtraction to avoid unnecessary damage to the effective speech signal.

[0092] Furthermore, in this embodiment, the estimated noise spectrum of each frame of the signal to be processed is obtained, the spectrum of all the signals to be processed and their estimated noise spectrum are used as inputs to spectral subtraction, the over-subtraction exponents of all the signals to be processed are used as over-detection factors in spectral subtraction, and the denoised signals to be processed are output.

[0093] It should be noted that noise spectrum estimation and spectral subtraction are well-known techniques, and the specific process of noise spectrum estimation and spectral subtraction for suppressing howling will not be elaborated here.

[0094] Thus, this embodiment constructs a spectral subtraction smoothing coefficient by fusing energy distribution characteristic values ​​with noise probability, and adaptively adjusts the over-subtraction exponent based on the dynamic changes of this coefficient between adjacent frames, thereby achieving real-time optimization of the spectral subtraction noise suppression intensity. This effectively suppresses residual feedback noise while preserving the original quality of the speech signal to the maximum extent, significantly improving the feedback suppression effect of digital hearing aids in complex acoustic environments.

[0095] Based on the same inventive concept as the above methods, this application also provides a digital hearing aid feedback suppression system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described digital hearing aid feedback suppression methods.

[0096] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0097] The various embodiments in the specification are described in progressive manner, and the same or similar parts between the various embodiments can be mutually referred to, and each embodiment focuses on the difference from other embodiments.

[0098] The above only describes the preferred embodiments of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for suppressing whistling in a digital hearing aid, characterized in that, The method includes the following steps: Acquire speech signals from a digital hearing aid and segment the speech signals into multiple signals to be processed; Based on the energy distribution of each signal to be processed in the frequency domain, the energy value of each signal to be processed in the frequency domain is divided into high and low energy values; based on the difference in the distribution between high and low energy values ​​of each signal to be processed in the frequency domain, and the degree of disorder of all energy values, the energy distribution characteristic value of each signal to be processed is determined. Based on the distribution of all high-energy values ​​and their derived values ​​in the frequency domain for each signal to be processed and a preset number of signals to be processed before it, a forward evaluation index for each signal to be processed is determined; based on the rate at which each peak in each signal to be processed drops to its next adjacent valley, an oscillation evaluation parameter for each signal to be processed is determined; based on the difference in oscillation evaluation parameters between each signal to be processed and a preset number of signals to be processed after it, a backward evaluation index for each signal to be processed is determined, and combined with the forward evaluation index, the noise probability of each signal to be processed is determined; Based on the energy distribution characteristic value and the noise probability, the over-attenuation exponent for each signal to be processed is determined; based on the over-attenuation exponent, spectral subtraction is used to suppress howling noise in the speech signal; The method for determining the energy distribution characteristic value of each signal to be processed is as follows: Calculate the proportion of high energy values ​​with frequencies greater than the fundamental frequency in the frequency domain for each signal to be processed to the total number of high energy values ​​in the signal to be processed, and denote it as the high energy distribution value; calculate the proportion of low energy values ​​with frequencies greater than the fundamental frequency in the frequency domain for each signal to be processed to the total number of low energy values ​​in the signal to be processed, and denote it as the low energy distribution value; Calculate the difference between the high-energy distribution value and the low-energy distribution value, and use the result of positively fusing the information entropy of all energy in the frequency domain of each signal to be processed with the difference as the energy distribution feature value of each signal to be processed; The expression for the over-decrease exponent of each signal to be processed is: In the formula , These represent the over-attenuation exponents of the i-th and (i-1)-th signals to be processed, respectively, where the initial value of the over-attenuation exponent is a preset value; , Let represent the spectral smoothing coefficients of the i-th and (i-1)-th signals to be processed, respectively. The spectral smoothing coefficient of each signal to be processed is the normalized value of the energy distribution characteristic value and the normalized mean of the noise probability. This indicates the preset adjustment step size.

2. The method for suppressing feedback in a digital hearing aid as described in claim 1, characterized in that, The step of dividing the energy value of each signal to be processed in the frequency domain into high and low energy values ​​includes: Calculate the mean of all energy values ​​of each signal to be processed in the frequency domain, and denote it as the global threshold. All energy values ​​of each signal to be processed that are greater than the global threshold are taken as high energy values, and the remaining energy values ​​are taken as low energy values.

3. The method for suppressing feedback in a digital hearing aid as described in claim 1, characterized in that, The method for determining the forward evaluation index of each signal to be processed is as follows: For each signal to be processed, obtain the mean, sum, maximum value, and number of all high-energy values ​​in the frequency domain; For each signal to be processed and the mean, sum of high energy values, maximum value, and number of high energy values ​​of a preset number of signals to be processed, four fitted lines are obtained. The sum of the slopes of the four fitted lines and the sum of the fitting errors are calculated respectively. The forward evaluation index of each signal to be processed is negatively correlated with the slope and value of its corresponding fitted line and the fitting error of the fitted line.

4. The method for suppressing feedback in a digital hearing aid as described in claim 1, characterized in that, The method for determining the oscillation evaluation parameters for each signal to be processed is as follows: The rate at which each peak in each signal to be processed drops to its next adjacent trough is denoted as the rate of decline of each peak. The average of the differences in the rates of decline between all peaks and their next adjacent peaks is used as the oscillation evaluation parameter for each signal to be processed.

5. The method for suppressing feedback in a digital hearing aid as described in claim 1, characterized in that, The backward evaluation index for each signal to be processed is the sum of the differences in oscillation evaluation parameters between each signal to be processed and a preset number of subsequent signals to be processed.

6. The method for suppressing feedback in a digital hearing aid as described in claim 1, characterized in that, The noise probability of each signal to be processed is the result of the positive fusion of the normalized value of the forward evaluation index and the normalized value of the backward evaluation index of each signal to be processed.

7. The method for suppressing feedback in a digital hearing aid as described in claim 1, characterized in that, The method of suppressing howling noise in speech signals using spectral subtraction includes: The estimated noise spectrum of each frame of the signal to be processed is obtained. The spectra of all the signals to be processed and their estimated noise spectra are used as inputs to the spectral subtraction method. The over-subtraction exponents of all the signals to be processed are used as over-detection factors in the spectral subtraction method. The denoised signals to be processed are then output.

8. A digital hearing aid feedback suppression system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the digital hearing aid feedback suppression method as described in any one of claims 1-7.

Images