Hearing aid fitting method and system

Abstract

The application relates to a hearing aid debugging method and system. By obtaining a reference parameter of a hearing aid and applying the reference parameter to the hearing aid, the hearing aid has an initial reference value before starting to adjust the parameter, so that the subsequent debugging process can be ensured not to deviate from the result, and the debugging efficiency is improved. The change of an emotional balance index and a psychological stress index of a person to be debugged when the person to be debugged wears the hearing aid for hearing test is taken as the basis for adjusting the parameter of the hearing aid, so that feedback can be quickly obtained without language communication with the person to be debugged, the emotional level and the psychological stress level of the person to be debugged when the person to be debugged wears the hearing aid can be known in real time, and the parameter of the hearing aid can be adjusted in time according to the information to improve the hearing feeling of the person to be debugged. In the adjustment process of the parameter of the hearing aid, coarse adjustment and fine adjustment are sequentially performed on the parameter of the hearing aid through two different step adjustment modes, so that the most comfortable area of the hearing feeling of the person to be debugged can be approached at the fastest speed, and the accuracy of the finally obtained parameter of the hearing aid is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of hearing aids, in particular to a hearing aid debugging method and system. BACKGROUND

[0002] The fitting process of a digital hearing aid is as follows: firstly, the hearing loss condition of a customer is determined by using pure tone audiometry, acoustic immittance, electrophysiology and the like; secondly, the parameters are preliminarily debugged by using an intelligent fitting algorithm; and finally, whether the hearing aid parameter debugging is appropriate is verified by using speech communication, sound field evaluation, speech evaluation, real ear test and the like.

[0003] The detection process is single in scene, strong in instantaneity and poor in continuity, the key steps such as pure tone audiometry are too subjective, and the above factors will introduce test errors, thereby leading to inaccurate debugging data; the verification process can calibrate the debugging data, but the feedback mechanism is too single, the elderly hearing loss customers are mostly not clear in expression, and it is difficult for them to cooperate with the fitter to complete accurate calibration. Therefore, the problem of multiple and repeated debugging still being unsatisfactory occurs, or the situation of good hearing in the fitting institution and poor use in daily life occurs.

[0004] In summary, the traditional hearing aid fitting process lacks real-time feedback steps and timely adjustment of hearing aid parameters according to user feedback, resulting in inaccurate hearing aid parameters.

[0005] To solve the above problems, a self-adaptive correction process needs to be added to the original fitting and verification process, and the application provides a self-adaptive hearing aid debugging method based on wearer emotional stress state evaluation, which can objectively evaluate the use of the user and automatically debug and optimize the data. SUMMARY

[0006] Therefore, it is necessary to provide a hearing aid debugging method and system to solve the problems of lack of real-time feedback steps and lack of timely adjustment of hearing aid parameters according to user feedback in the traditional hearing aid fitting process.

[0007] The application provides a hearing aid debugging method, which comprises the following steps:

[0008] obtaining reference parameters of a hearing aid and applying the reference parameters of the hearing aid to the hearing aid;

[0009] increasing the hearing aid parameters by a first preset percentage of the current value, and applying the adjusted hearing aid parameters to the hearing aid and performing a hearing test on the sample;

[0010] obtaining an emotional balance index and a psychological stress index of the sample in real time during the hearing test on the sample, and determining whether the emotional balance index and the psychological stress index increase during the hearing test;

[0011] If the emotional balance index and the psychological stress index are both increased during the hearing test, the hearing aid parameter is adjusted by a first preset percentage of the current value, and the adjusted hearing aid parameter is applied to the hearing aid and the hearing test is performed on the sample again;

[0012] If the emotional balance index and the psychological stress index are both decreased during the hearing test, the hearing aid parameter is adjusted by a second preset percentage of the current value, and the adjusted hearing aid parameter is applied to the hearing aid and the hearing test is performed on the sample again; the second preset percentage is less than the first preset percentage;

[0013] The emotional balance index and the psychological stress index of the sample are obtained in real time during the hearing test, and it is determined whether the emotional balance index and the psychological stress index are both increased during the hearing test;

[0014] If the emotional balance index and the psychological stress index are both increased during the hearing test, the hearing aid parameter is adjusted by a second preset percentage of the current value, and the adjusted hearing aid parameter is applied to the hearing aid and the hearing test is performed on the sample again;

[0015] If the emotional balance index and the psychological stress index are both decreased during the hearing test, the current value of the hearing aid is outputted;

[0016] The current value of the hearing aid is written into the hearing aid chip of the hearing aid as the optimal hearing aid parameter.

[0017] The application also provides a hearing aid debugging system, comprising:

[0018] A cloud server is configured to perform the hearing aid debugging method as mentioned in the foregoing content;

[0019] A hearing aid is in communication connection with the cloud server.

[0020] The application relates to a hearing aid debugging method and system. First, the reference parameters of a hearing aid are acquired, and the reference parameters of the hearing aid are applied to the hearing aid, so that the hearing aid has an initial reference value before starting to adjust parameters, and the reference value can ensure that the subsequent debugging process will not deviate from the result, thereby improving the debugging efficiency. Second, the change of the emotional balance index and the psychological stress index of a person to be debugged when the person to be debugged wears the hearing aid for hearing test is taken as the basis for adjusting the parameters of the hearing aid, so that the feedback of the person to be debugged can be quickly obtained without language communication with the person to be debugged, and the feedback is the emotional balance index and the psychological stress index. Therefore, the scheme of the application can know the emotional level and the psychological stress level of the person to be debugged when the person to be debugged wears the hearing aid in real time and quickly according to the emotional balance index and the psychological stress index, and the parameters of the hearing aid can be adjusted in time according to the information to improve the hearing feeling of the person to be debugged. Finally, in the process of adjusting the parameters of the hearing aid, the parameters of the hearing aid are coarsely adjusted and finely adjusted through two different step amplitudes, specifically, the parameters of the hearing aid are increased by a first preset percentage of a larger value step until the emotional balance index and the psychological stress index of the person to be debugged are monitored to decrease, and then the parameters of the hearing aid are decreased by a second preset percentage of a smaller value step until the emotional balance index and the psychological stress index of the person to be debugged decrease, and finally the optimal hearing aid parameters are output. The improved hearing aid debugging method can quickly approach the most comfortable area of the hearing feeling of the person to be debugged, so that the final optimal hearing aid parameters have high accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A flowchart of a hearing aid debugging method provided by an embodiment of the application is shown.

[0022] Figure 2 A structural diagram of a hearing aid debugging system provided by an embodiment of the application is shown.

[0023] Reference signs:

[0024] 100 - cloud server; 200 - hearing aid. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the application and do not limit the application.

[0026] The application provides a hearing aid debugging method. It should be noted that the hearing aid debugging method provided by the application is applied to any kind, brand and appearance of hearing aid.

[0027] Further, the hearing aid debugging method provided in the application is not limited in its execution subject. Alternatively, the execution subject of the hearing aid debugging method provided in the application can be a hearing aid debugging system. Specifically, the execution subject of the hearing aid debugging method provided in the application can be a cloud server 100 in the hearing aid debugging system.

[0028] As shown in the embodiment of the application, the hearing aid debugging method comprises the following S100 to S800: Figure 1

[0029] S100, obtaining a reference parameter of a hearing aid, and applying the reference parameter of the hearing aid to the hearing aid.

[0030] S200, increasing the hearing aid parameter by a first preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample.

[0031] S300, obtaining the emotional balance index and the psychological stress index of the sample in real time during the hearing test, and determining whether the emotional balance index and the psychological stress index during the hearing test are both increased.

[0032] S400, if the emotional balance index and the psychological stress index during the hearing test are both increased, returning to the step of increasing the hearing aid parameter by the first preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample.

[0033] S500, if the emotional balance index and the psychological stress index during the hearing test are both decreased, decreasing the hearing aid parameter by a second preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample; the second preset percentage is smaller than the first preset percentage.

[0034] S600, obtaining the emotional balance index and the psychological stress index of the sample in real time during the hearing test, and determining whether the emotional balance index and the psychological stress index during the hearing test are both increased.

[0035] S700, if the emotional balance index and the psychological stress index during the hearing test are both increased, returning to the step of decreasing the hearing aid parameter by the second preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample.

[0036] S800, if the emotional balance index and the psychological stress index during the hearing test are both decreased, outputting the current value of the hearing aid.

[0037] S900, writing the current value of the hearing aid as the optimal hearing aid parameter into the hearing aid chip of the hearing aid.

[0038] ​Specifically, in the present application, the sample refers to the person being fitted.

[0039] In S100, the reference parameter is the benchmark parameter for adjusting the hearing aid parameters. Without the reference parameter, the subsequent parameter adjustment has no meaning and is easy to deviate from the true hearing comfort zone of the sample, reducing the fitting efficiency.

[0040] S200 to S400 are the coarse adjustment process. It can be understood that in the coarse adjustment process, we use a first preset percentage that is larger than the second preset percentage. Optionally, the first preset percentage can be 10%.

[0041] S500 to S800 are the fine adjustment process. It can be understood that in the fine adjustment process, we use a second preset percentage that is smaller than the first preset percentage. Optionally, the first preset percentage can be 5%.

[0042] Optionally, the current value of the hearing aid output in S800 can be written as the optimal hearing aid parameter into the hearing aid chip of the hearing aid.

[0043] The following details what the hearing aid parameter to be increased or decreased is.

[0044] The hearing aid parameter is not a single parameter, but consists of many complex parameters. Different types of hearing aids and different brands of hearing aids have different parameters included in the hearing aid parameter.

[0045] The hearing aid parameter mainly includes two categories of parameters, one is the hearing gain parameter, and the other is the noise reduction parameter. The main function of the hearing gain parameter is to amplify the sound signal and compensate for the hearing loss of the user. The noise reduction parameter is a modification of the noise.

[0046] Optionally, the present application can perform S100 to S800 on the hearing gain parameter once, determine the optimal parameter of the hearing gain parameter, and write the optimal parameter of the hearing gain parameter into the hearing aid chip. After that, the noise reduction parameter is executed again S100 to 800, so as to obtain the optimal hearing aid parameter.

[0047] Optionally, the hearing aid parameter includes but is not limited to the hearing gain in each frequency band, the noise reduction in each frequency band, the strength of feedback suppression, and WDRC.

[0048] The hearing gain in each frequency band (Frequency-Specific Gain) refers to the amplification capability of the hearing aid to the sound signal at different frequencies (such as low frequency, medium frequency, and high frequency), which aims to compensate for the hearing loss of the user in different frequency bands. For example, a user with high-frequency hearing loss needs higher high-frequency gain. The unit of hearing gain is decibel (dB).

[0049] Noise Reduction by Frequency Band refers to the ability of a hearing aid to suppress background noise in different frequency bands. Hearing aids can identify and reduce non-speech noise (e.g., fan noise, traffic noise) in different frequency bands through algorithms while preserving speech signals. Noise reduction can be measured in decibels (dB) or percentage (%). When measured in dB, noise reduction refers to the absolute reduction in noise, such as 15 dB. When measured in percentage, noise reduction refers to the intensity level of noise suppression, such as 80% representing the maximum noise suppression capability. Noise reduction can improve the signal-to-noise ratio of a hearing aid and improve speech intelligibility when the recipient is in a noisy environment.

[0050] Feedback suppression intensity refers to the ability to prevent a hearing aid from producing feedback whistling due to sound feedback. Feedback suppression can counteract signal oscillation in the feedback path (e.g., ear mold leakage). Feedback suppression can be measured in levels (low, medium, high) or decibels (dB). When measured in levels, feedback suppression represents the aggressiveness of the suppression algorithm. When measured in dB, feedback suppression accurately represents the gain reduction of the system at feedback frequencies.

[0051] WDRC (Wide Dynamic Range Compression) is a nonlinear amplification technique that provides high gain for weak sounds and low gain for strong sounds, allowing sounds of different intensities to fall within the recipient's audible dynamic range.

[0052] Sub-parameters and units of WDRC:

[0053] Compression Ratio: The ratio of input / output intensity (e.g., 2:1), with no unit.

[0054] Thres hold: The sound pressure level that triggers compression, measured in dB SPL (e.g., 40 dB SPL).

[0055] Attack Time: The delay before compression begins after detecting a strong signal, measured in milliseconds (ms), such as 5 ms.

[0056] Release Time: The delay before resuming amplification after the signal weakens, measured in milliseconds (ms), such as 100 ms.

[0057] WDRC can improve the audibility of weak sounds while avoiding discomfort for the recipient caused by strong sounds.

[0058] The hearing aid parameters directly affect the hearing aid effect. The hearing aid parameters with appropriate values can significantly improve the hearing experience of the person being tested, and improve the emotional balance index and psychological stress index of the person being tested. On the contrary, unreasonable hearing aid parameters can cause various uncomfortable situations of the person being tested, make the person being tested have a resistance to the hearing aid, and reduce the emotional balance index and psychological stress index of the person being tested, for example, the person being tested cannot hear clearly, the person being tested cannot communicate with the test sound with interactive nature, the person being tested feels too noisy, the person being tested feels that the noise is too loud, and the like.

[0059] In the embodiment, first, the reference parameters of the hearing aid are acquired, and the reference parameters of the hearing aid are applied to the hearing aid, so that the hearing aid has an initial reference value before starting to adjust the parameters. The reference value can ensure that the subsequent debugging process will not deviate from the result, and improve the debugging efficiency. Second, the changes of the emotional balance index and the psychological stress index of the person being tested when wearing the hearing aid for hearing test are taken as the basis for adjusting the hearing aid parameters. The feedback of the person being tested can be quickly obtained without language communication with the person being tested, and the feedback obtained is the emotional balance index and the psychological stress index. Therefore, the scheme of the present application can know the emotional level and the psychological stress level of the person being tested when wearing the hearing aid in real time and quickly according to the emotional balance index and the psychological stress index, and can timely adjust the hearing aid parameters according to the information to improve the hearing experience of the person being tested. Finally, in the adjustment process of the hearing aid parameters, the hearing aid parameters are coarsely adjusted and finely adjusted by two different step amplitudes. Specifically, the hearing aid parameters are increased by a first preset percentage of a larger value until the emotional balance index and the psychological stress index of the person being tested are monitored to decrease, and then the hearing aid parameters are decreased by a second preset percentage of a smaller value until the emotional balance index and the psychological stress index of the person being tested are monitored to decrease, and finally the optimal hearing aid parameters are output. The improved hearing aid debugging method can quickly approach the most comfortable area of the hearing experience of the person being tested, so that the final optimal hearing aid parameters have high accuracy.

[0060] In an embodiment of the present application, before S100, that is, before the reference parameters of the hearing aid are acquired and the reference parameters of the hearing aid are applied to the hearing aid, the hearing aid debugging method further includes the following S010 to S020:

[0061] S010, it is judged whether the hearing aid is first debugged.

[0062] S020, if the hearing aid is first debugged, the fitter debugging parameters are called, and a preset percentage of the fitter debugging parameters is taken as the reference parameters of the hearing aid.

[0063] Specifically, before knowing the hearing aid debugging method described in the application, an audiologist debugging parameter library can be established in advance. The audiologist debugging parameter library stores user portrait files and audiologist debugging parameters corresponding to the user portrait files. The audiologist debugging parameter is an initial hearing aid parameter generated by the hearing aid audiologist after individual adjustment of various functional parameters of the hearing aid according to the hearing loss condition of the user and the user's wearing experience. It is an initial hearing aid parameter generated by a traditional fitting method. There are many traditional fitting methods (see CN114827861A and CN205987370U). The fitting method is not the focus of the protection of the application, so it will not be explained in detail here. The user portrait file includes the user's age, the user's gender, the user's hearing loss condition data, etc.

[0064] When the audiologist debugging parameter is retrieved, all user portrait files in the audiologist debugging parameter library can be traversed to search for a user portrait file with the highest similarity to the sample user portrait file as an approximate user portrait file, and finally the audiologist debugging parameter corresponding to the approximate user portrait file is retrieved as the audiologist debugging parameter of the sample. There are many methods for calculating similarity. Optionally, the weights of each sub-parameter of the user's age, the user's gender, and the user's hearing loss condition data can be assigned, and then the user's age is compared and then normalized to the [0, 1] interval. If the user's gender is the same, it is normalized to 0, and if the user's gender is different, it is normalized to 1. Each sub-parameter of the user's hearing loss condition data is compared and then normalized to the [0, 1] interval. Finally, the weight of each data is multiplied by the normalized value, and the sum of the weight multiplied by the normalized value of each data is taken as the similarity value. The smaller the similarity value, the smaller the gap. The larger the similarity value, the larger the gap. We find the user portrait file with the smallest similarity value as the approximate user portrait file, and obtain the audiologist debugging parameter corresponding to the approximate user portrait file as the audiologist debugging parameter of the sample. Further, a predetermined percentage of the audiologist debugging parameter of the sample is taken as the reference parameter of the hearing aid.

[0065] Before S100, that is, before the reference parameter of the hearing aid is obtained and applied to the hearing aid, the hearing aid debugging method further comprises:

[0066] S030, if the hearing aid is not initially debugged, S100 is directly executed.

[0067] In this embodiment, by obtaining the reference parameter of the hearing aid and applying the reference parameter of the hearing aid to the hearing aid, the hearing aid has an initial reference value before starting to adjust the parameter, which can ensure that the result does not deviate during subsequent debugging and improve the debugging efficiency.

[0068] In an embodiment of the present application, the preset percentage is 80%.

[0069] Specifically, the preset percentage can also be any value greater than or equal to 50% and less than or equal to 90%. The preset percentage can be 50%. The preset percentage can be 90%. The preset percentage can be 80%. 80% is a relatively appropriate value.

[0070] In this embodiment, by setting 80% of the audiologist's debugging parameters as the reference parameter, a non-small adjustment space can be provided for subsequent adjustment of the hearing aid parameters, and the adjustment space will not be too large to waste time. This setting can maximize the adjustment efficiency of the hearing aid parameters.

[0071] In an embodiment of the present application, S800 includes, that is, the output of the current value of the hearing aid includes the following S801:

[0072] S801, the current value of the hearing aid is increased by a third preset percentage of the amplitude of the current value of the hearing aid, and the adjusted hearing aid parameter is output.

[0073] Specifically, the user's feeling is relatively subjective, and although the present application quantifies the user's hearing feeling from the emotional balance index and the psychological stress index, in order to avoid floating and errors, after outputting S800 to obtain the current value of the hearing aid, the current parameter is also increased by a certain percentage, which is the third preset percentage of the current value. This can give a certain redundancy interval. The adjusted hearing aid parameter is used as the new current value of the hearing aid to execute S900, that is, the adjusted hearing aid parameter is used as the optimal hearing aid parameter.

[0074] In this embodiment, by increasing the current value of the hearing aid by a third preset percentage of the amplitude of the current value of the hearing aid before obtaining the adjusted optimal hearing aid parameter, the hearing aid parameter can be further optimized, so that the hearing aid parameter has high redundancy in the subsequent application process, so that the hearing aid debugging method provided by the present application has reproducibility. According to tests, the deviation of the hearing aid parameters obtained by executing the hearing aid debugging method provided by the present application on the same person multiple times is not more than ±1%, and the stability is high.

[0075] In an embodiment of the present application, after S800 and before S900, that is, after the output of the current value of the hearing aid and before the current value of the hearing aid is written into the hearing aid chip of the hearing aid as the optimal hearing aid parameter, the hearing aid debugging method further includes the following S810 to S861b:

[0076] S810, create a monitoring number, and the initial value of the monitoring number is 0.

[0077] S820, taking the current value of the hearing aid as a benchmark parameter.

[0078] S830, setting a benchmark parameter fluctuation range. The benchmark parameter fluctuation range is [benchmark parameter - benchmark parameter x fourth preset percentage, benchmark parameter + benchmark parameter x fourth preset percentage].

[0079] S840, performing fluctuation test on the current value of the hearing aid, applying the current value of the hearing aid to the hearing aid and performing hearing test on the sample.

[0080] S850, obtaining the emotional balance index and the psychological stress index of the sample in real time when performing the hearing test on the sample, and determining whether the emotional balance index and the psychological stress index of the sample during the hearing test are within the standard parameter fluctuation range.

[0081] S861, if the emotional balance index and the psychological stress index of the sample during the hearing test are within the standard parameter fluctuation range, increasing the monitoring number by 1 based on the original value, and determining whether the current value of the monitoring number is greater than or equal to the preset monitoring number.

[0082] S861a, if the current value of the monitoring number is less than the preset monitoring number, returning to S840 after a preset time period, i.e. returning to the fluctuation test on the current value of the hearing aid, applying the current value of the hearing aid to the hearing aid and performing hearing test on the sample.

[0083] S861b, if the value of the monitoring number is greater than or equal to the preset monitoring number, adjusting the value of the monitoring number to 0, and outputting the current value of the hearing aid.

[0084] Specifically, the current value of the hearing aid obtained in S800 is not the optimal hearing aid parameter, but the fluctuation degree thereof needs to be monitored, and whether it is the optimal hearing aid parameter is determined according to the monitoring result. S510 to S861b is a complete process of monitoring the fluctuation degree.

[0085] In S830, the fourth preset percentage can be 5%. The benchmark parameter fluctuation range is [95% of the benchmark parameter, 105% of the benchmark parameter], i.e. we need to monitor the fluctuation of the current value of the hearing aid, if it does not exceed the range of ±5%, it indicates that the current value of the hearing aid is qualified, and further determining whether the monitoring number is greater than or equal to the preset monitoring number.

[0086] The preset monitoring number can be 10.

[0087] In one embodiment, S810 is performed after S800, i.e. the current value of the hearing aid outputted in S800 is taken as the benchmark parameter in the subsequent S820, and then the subsequent steps S820-S861b are performed.

[0088] In one embodiment, S810 is performed after S801, i.e. the adjusted hearing aid parameter outputted in S801 is taken as the benchmark parameter in the subsequent S820, and then the subsequent steps S820-S861b are performed.

[0089] S900 is performed after S861b, i.e. the current value of the hearing aid in S861b in the present embodiment is taken as the optimal hearing aid parameter.

[0090] The preset time period can be 10 seconds.

[0091] In the present embodiment, after obtaining the current value of the hearing aid, the current value of the hearing aid is further subjected to multiple fluctuation tests, and only when the results of the continuous multiple fluctuation tests are qualified can the current value of the hearing aid be taken as the optimal hearing aid parameter. Through such fluctuation degree test, the optimal hearing aid parameter with higher accuracy can be obtained, and some non-objective interference factors such as hearing aid wearing not in accordance with regulations, sound leakage, and the first participation in the debugging by the person being debugged can be excluded. In addition, a time interval, i.e. a preset time period, is set between the adjacent two fluctuation tests, so as to avoid the problem of large change in psychological pressure of the person being debugged caused by continuous multiple tests, and further improve the effectiveness of the fluctuation test.

[0092] In one embodiment of the present application, S850 is further included, i.e. the emotional balance index and the psychological pressure index of the sample are acquired in real time when the sample is subjected to the hearing test, and after it is judged whether the emotional balance index and the psychological pressure index of the sample during the hearing test are located within the standard parameter fluctuation range, the following S862 is further included:

[0093] S862, if the emotional balance index and the psychological pressure index of the sample during the hearing test are not located within the standard parameter fluctuation range, the value of the monitoring number is adjusted to 0, and the hearing aid parameter is adjusted by the first preset percentage of the amplitude of the current value, and the adjusted parameter is applied to the hearing aid and the sample is subjected to the hearing test.

[0094] Specifically, the present embodiment requires that the optimal hearing aid parameter is outputted only when the sample continuously does not exceed the benchmark parameter fluctuation range for multiple times, and the number of times is the preset monitoring number, so in S862 we can see that when the emotional balance index and the psychological pressure index of the sample during the hearing test are not located within the standard parameter fluctuation range, the value of the monitoring number is reset to 0, and the debugging is restarted from S200.

[0095] In the embodiment, the number of monitoring times is adjusted to 0 when the emotional balance index and the psychological stress index of the sample during the hearing test are not within the standard parameter fluctuation range, ensuring the continuity of the fluctuation degree test qualification, improving the stability of the optimal hearing aid parameters, and only when the hearing aid parameters do not deviate from the fluctuation for a plurality of times, the optimal hearing aid parameters are considered.

[0096] In an embodiment of the present application, the hearing aid debugging method further comprises creating a cycle number, and the initial value of the cycle number is 0.

[0097] After S861b, the following S871-S878b are further included, i.e., after judging whether the current value of the monitoring number is greater than or equal to the preset monitoring number, the following S871-S878b are further included:

[0098] S871, the current hearing aid parameter is taken as a new reference parameter.

[0099] S872, the cycle number is increased by 1 based on the original value.

[0100] S873, judging whether the current value of the cycle number is greater than or equal to the preset cycle number.

[0101] S874a, if the current value of the cycle number is less than the preset cycle number, returning to the reference parameter of the hearing aid, and applying the reference parameter of the hearing aid to the hearing aid.

[0102] S874b, if the current value of the cycle number is greater than or equal to the preset cycle number, obtaining the reference parameter obtained after each cycle.

[0103] S875, sorting the reference parameters obtained after each cycle in descending order, and selecting the maximum reference parameter and the minimum reference parameter.

[0104] S876, calculating the difference between the maximum reference parameter and the minimum reference parameter, and calculating the percentage of the difference in the minimum reference parameter, and defining the percentage as a fluctuation percentage.

[0105] S877, judging whether the fluctuation percentage is less than or equal to the preset fluctuation percentage.

[0106] S878a, if the fluctuation percentage is less than or equal to the preset fluctuation percentage, the value of the cycle number is adjusted to 0, the average value of the reference parameters obtained after each cycle is calculated, and the average value is taken as the optimal hearing aid parameter.

[0107] S878b, writing the optimal hearing aid parameter into the hearing aid chip of the hearing aid.

[0108] Specifically, the concept of "cycle" is introduced in the present embodiment. If the current value of the hearing aid obtained in S800 does not exceed the benchmark parameter fluctuation range for 10 times in succession, the complete process of monitoring the fluctuation degree ends, i.e. one cycle ends, according to the foregoing S810 to S862 embodiment.

[0109] In the foregoing S810 to S862 embodiment, the current value of the hearing aid is taken as the optimal hearing aid parameter after S861b is executed, i.e. the current value of the hearing aid is within the benchmark parameter fluctuation range for 10 times in succession, and the current value of the hearing aid is taken as the optimal hearing aid parameter. However, in the present embodiment, the current value of the hearing aid needs to be further tested for the fluctuation degree.

[0110] In the present embodiment, S861b is executed, and then it is considered that one cycle ends. The present embodiment also needs to execute the same cycle multiple times, and the specific number of times is the preset cycle number. Since a relatively optimal hearing aid parameter is finally obtained after each cycle, the present embodiment and the foregoing embodiment take the relatively optimal hearing aid parameter as the reference parameter for the next cycle and re-cycle. Due to errors, fluctuations may still occur. The present embodiment continues to execute the next cycle after one cycle ends, and summarizes the overall monitoring result after executing all the cycles of the preset cycle number. The summary is a summary of the fluctuation of multiple cycles, and the summary of the present embodiment also needs to ensure the continuity of qualified data, so when the fluctuation percentage is less than or equal to the preset fluctuation percentage, the value of the cycle number is set to 0, the previous cycles are discarded, and the first cycle is started again. Until the fluctuation percentages of multiple cycles (the number of times is the preset cycle number) are all less than or equal to the preset fluctuation percentage, the average value of the reference parameters obtained after each cycle is calculated, and the average value is taken as the optimal hearing aid parameter.

[0111] Optionally, the preset cycle number can be 5.

[0112] In the present embodiment, the current hearing aid parameter obtained after one cycle ends is taken as the reference parameter used when the next cycle starts. After the cycles of the prediction cycle number are executed, the fluctuation percentage is calculated based on the current hearing aid parameter obtained after each cycle, and the fluctuation percentage is compared with the preset percentage, so that the stability of the hearing aid parameter in multiple cycles can be controlled.

[0113] Optionally, the fluctuation percentage can be greater than a fourth preset percentage, and the fourth preset percentage can be 5% and the fluctuation percentage can be 10%. It can be understood that the numerical adjustment of the hearing aid parameter is coarse adjustment first and then fine adjustment, and the fluctuation degree monitoring in the embodiment is fine monitoring first and then coarse monitoring, that is, the fluctuation degree monitoring in a single cycle uses the fourth preset percentage with a smaller value, and the fluctuation degree monitoring after a plurality of cycles uses the fluctuation percentage with a larger value, so that the fluctuation degree can be controlled to be minimized in a single cycle, and the fluctuation percentage is relatively more relaxed in a plurality of cycles. The sensitivity of the whole fluctuation monitoring process is just right, and the efficiency of the fluctuation degree monitoring is improved.

[0114] In an embodiment of the present application, after S877, that is, after determining whether the fluctuation percentage is less than or equal to the preset fluctuation percentage, S879 is further included:

[0115] S879, if the fluctuation percentage is greater than the preset fluctuation percentage, the value of the cycle number is adjusted to 0, the reference parameter of the hearing aid is obtained, and the reference parameter of the hearing aid is applied to the hearing aid.

[0116] Specifically, the principle of step S879 is similar to that of S862, which will not be described here.

[0117] In an embodiment of the present application, after S900, that is, after the current value of the hearing aid is written into the hearing aid chip of the hearing aid as the optimal hearing aid parameter, the hearing aid debugging method further includes:

[0118] S910, generating a personal archive of the sample, and storing the optimal hearing aid parameter corresponding to the ID of the sample into the personal archive of the sample.

[0119] Specifically, the optimal hearing aid parameter of each sample ID in the personal archive of the sample is not fixed, and the optimal hearing aid parameter can be invalid due to the changes of the age of the person to be debugged, the working environment, and the service life of the hearing aid.

[0120] Therefore, in an embodiment of the present application, the hearing aid debugging method mentioned in the present application is performed on the hearing aid again every certain period of time, and the period of time can be any one of one month to three years. For example, the hearing aid debugging method mentioned in the present application is used to debug the hearing aid once every three months, and the personal archive of the sample is updated after debugging. It should be noted that when S100 is performed, the optimal hearing aid parameter of the sample ID in the personal archive of the sample can be retrieved as the reference parameter of the hearing aid.

[0121] In this embodiment, the optimal hearing aid parameters are stored in the personal profile of the sample by generating the personal profile of the sample, so as to facilitate subsequent analysis and traceability.

[0122] The optimal hearing aid parameters obtained by the hearing aid debugging method provided in the application have objectivity and stability, so that the hearing aid debugging method provided in the application has reproducibility, and the fluctuation of the results after repeated implementation is extremely small. It is tested that the overall deviation degree of the hearing aid parameters finally obtained by executing the hearing aid debugging method provided in the application on the same person to be debugged for multiple times is not more than ±5%. A hearing-impaired user with an age of 50 years old and a gender of male is repeatedly taken as the person to be debugged to perform the hearing aid debugging method provided in the application for multiple times, and the debugging results are shown in Table 1 (only the first three debugging results are shown due to the limitation of the length).

[0123] Table 1- hearing aid parameter debugging result table

[0124]

[0125] In an embodiment of the application, the hearing aid debugging method provided in the application further comprises:

[0126] W100, collecting the vital sign data of the sample by using the test sound, and collecting the expression data when the vital sign data is collected.

[0127] W200, calling the hearing evaluation data of the sample.

[0128] W300, creating a heart rate variability prediction model.

[0129] W400, inputting the vital sign data of the sample and the hearing evaluation data as training data into the heart rate variability prediction model to train the heart rate variability prediction model. The output data of the heart rate variability prediction model is the heart rate variability.

[0130] W500, creating an emotion and psychological evaluation model.

[0131] W600, inputting the output data of the heart rate variability prediction model, the hearing evaluation data, and the expression data when the vital sign data of the sample is collected as training data into the emotion and psychological evaluation model to train the emotion and psychological evaluation model. The output data of the emotion and psychological evaluation model is the emotion balance index and the psychological stress index.

[0132] In an embodiment of the application, the W100 comprises:

[0133] W110, playing the test sound, and collecting the heart rate sequence of the sample in the process of playing the test sound.

[0134] Specifically, the test sound can be a 5-minute audio file, and the audio file can be pure music. The heart rate is collected every 5 seconds, and 60 heart rates can be obtained, which are arranged in the order of the collection time nodes to form a heart rate sequence. The heart rate sequence has 60 heart rates. For example, [72, 75, 71,..., 68], unit: bpm, times / minute.

[0135] In an embodiment of the present application, the W100 further comprises:

[0136] W120, when collecting the heart rate sequence of the sample, the expression data of the sample is synchronously collected.

[0137] Specifically, a camera can be arranged in front of the sample, and the arrangement position of the camera needs to be such that the image capturing range of the camera completely covers the face of the sample. When the heart rate is collected every 5 seconds, the camera also synchronously captures the face image of the sample. Subsequently, 60 face images can also be obtained. The analysis module inside the camera can identify and analyze the 60 face images, and obtain the expression data of the sample after identification and analysis. The analysis module inside the camera pre-models the face template of the sample, and marks the key points in the face. The key points in the face are the key points in the outline of the facial features and the key points in the outline of the face, and the key points in the face include but are not limited to 68 key points such as left eyebrow peak, left eyebrow center, left eyebrow tail, right eyebrow peak, right eyebrow center, right eyebrow tail, left pupil, right pupil, left inner corner of eye, left outer corner of eye, right inner corner of eye, right outer corner of eye, nose tip, nose root, left mouth corner, right mouth corner, upper lip center point, left cheekbone, and right cheekbone.

[0138] The analysis module can identify the key points of the face and fuse them with the two-dimensional rectangular coordinate system to calculate the coordinates of each key point. The analysis module calculates the expression data according to the coordinates of each key point.

[0139] For example, the expression data includes the vertical offset of the mouth corner, the pupil distance change rate, and the eyebrow distance.

[0140] Method for calculating the vertical offset of the mouth corner:

[0141] The vertical coordinates of the upper lip center point, the left mouth corner, and the right mouth corner are obtained.

[0142] The vertical offset of the mouth corner is calculated according to Formula 1.

[0143] Formula 1.

[0144] wherein, is the vertical offset, is the vertical coordinate of the upper lip center point, is the vertical coordinate of the left mouth corner, is the vertical coordinate of the right corner of the mouth.

[0145] The method for calculating the pupil distance change rate is as follows:

[0146] The pupil distance change rate is calculated according to formula 2.

[0147] Formula 2.

[0148] wherein, is the pupil distance change rate, is the baseline pupil distance, is the current pupil distance. The pupil distance is the horizontal distance between the pupils of the two eyes, which is calculated according to the absolute value of the difference between the horizontal coordinate of the left pupil and the horizontal coordinate of the right pupil. The baseline pupil distance is the pupil distance of the user in a calm state.

[0149] The inter-brow distance can be calculated according to the absolute value of the difference between the horizontal coordinate of the left brow tail and the horizontal coordinate of the right brow tail.

[0150] In an embodiment of the present application, the W210 comprises:

[0151] The W210 retrieves the low-frequency threshold, the medium-frequency threshold and the high-frequency threshold of the sample.

[0152] Specifically, the hearing evaluation data of the sample in the present embodiment uses the low-frequency threshold, the medium-frequency threshold and the high-frequency threshold. The collection method is pure tone audiometry, that is, the user presses the response button when the user perceives the sound. The measured hearing evaluation data is stored as the inherent property of the sample in the personal file of the sample.

[0153] The low-frequency threshold is the minimum sound intensity that the user can perceive at a frequency of 250 Hz, and the unit is dB HL. The medium-frequency threshold is the minimum sound intensity that the user can perceive at a frequency of 1 kHz, and the unit is dB HL. The high-frequency threshold is the minimum sound intensity that the user can perceive at a frequency of 4 kHz, and the unit is dB HL.

[0154] Before the W300, it further comprises:

[0155] The W220 performs cleaning processing on the physical data of the sample, the expression data when the physical data is collected, and the hearing evaluation data.

[0156] Specifically, the cleaning of the physical data can be performed through several screening conditions. For example, an abnormal heart rate judgment is performed, and specifically, each heart rate single-point data is traversed to judge whether each heart rate single-point data satisfies the following two conditions at the same time:

[0157] Condition 1: The heart rate single-point data is located in the range of [30 times / minute, 180 times / minute].

[0158] Condition 2: The change rate of the continuous three heart rate single-point data is less than 20%.

[0159] If a single point of heart rate data meets the following two conditions at the same time, it is determined that the single point of heart rate data is qualified.

[0160] If a single point of heart rate data does not meet at least one condition, it is determined that the single point of heart rate data is unqualified. That is, condition 1 is not met, or condition 2 is not met, or both condition 1 and condition 2 are not met, then the single point of heart rate data is unqualified.

[0161] Optionally, linear interpolation processing is performed on the unqualified single point of heart rate data to replace the unqualified single point of heart rate data. Optionally, in a specific scheme, the average value of the two qualified points before and after the unqualified single point of heart rate data is calculated to replace the unqualified single point of heart rate data.

[0162] The cleaning of the hearing evaluation data can be completed in the following way:

[0163] Determine whether there is any one frequency band whose frequency band threshold is greater than 120 dB. If there is any one frequency band whose frequency band threshold is greater than 120 dB, mark the frequency band threshold greater than 120 dB as an invalid sample, and directly eliminate the frequency band threshold.

[0164] W230, the normalized processing is performed on the cleaned physical sign data, the expression data collected when the cleaned physical sign data, and the cleaned hearing evaluation data.

[0165] Specifically, the normalized processing is performed on the physical sign data according to formula 3. In an embodiment of the present application, the normalized processing is performed on the heart rate.

[0166] Formula 3.

[0167] Wherein, is the normalized heart rate, is the heart rate to be normalized, is the average value of all heart rates in the heart rate sequence, is the standard deviation of all heart rates in the heart rate sequence.

[0168] In an embodiment of the present application, the normalized processing is performed on the low frequency threshold, the medium frequency threshold and the high frequency threshold according to formula 4.

[0169] Formula 4.

[0170] Wherein, is the normalized threshold, is the threshold to be normalized, is the maximum hearing loss.

[0171] For example, the maximum hearing loss is 120 dB, and the low-frequency threshold of the person being tested is 60 dB, so the normalized low-frequency threshold is 0.5. The normalization of the medium-frequency threshold and the high-frequency threshold is the same.

[0172] The following describes the normalization method of expression data.

[0173] The normalization of the mouth corner vertical offset can be normalized by formula 5, which quantifies the mouth corner score.

[0174] Formula 5.

[0175] Wherein, is the mouth corner score. is the mouth corner vertical offset. is the vertical offset reference value, which is determined by a large number of samples.

[0176] When in a natural state, ≈0. When smiling, >0. The greater the absolute value of the mouth corner score, the higher the intensity of the mouth corner lifting. When sad or dissatisfied, <0, The greater the absolute value of the mouth corner score, the higher the intensity of the mouth corner lifting. Then after normalization, is mapped to the interval [-1, 1] to obtain the mouth corner score, and the mouth corner score The maximum value of the mouth corner score is 1, and the minimum value of the mouth corner score The closer to 1, the closer to the smiling state, The closer to -1, the closer to the sad or dissatisfied state, is 0, which is in a natural state. The pupil distance change rate itself is a ratio close to 1 and does not need to be normalized.

[0177] The normalization of the inter-brow distance uses the ratio of the inter-brow distance to the face width to calculate, for example, the inter-brow distance is 35 mm, and the face width is 140 mm, the ratio of the two is 0.25, and 0.25 is the normalized inter-brow distance. The face width can be set as the straight line distance between the left and right zygomatic bone coordinate points, which is obtained by calculating the difference between the horizontal coordinates of the left and right zygomatic bone coordinate points.

[0178] W400 includes, that is, inputting the physical data and hearing evaluation data of the sample into the heart rate variability prediction model as training data, training the heart rate variability prediction model, including:

[0179] W410, data fusion is performed on the training data.

[0180] W410, data fusion is performed on the training data.

[0181] W410 includes:

[0182] W411 extracts the heart rate time-dependent feature using the LSTM network, as shown in Equation 6.

[0183] Equation 6.

[0184] wherein, is the hidden state of the current time node t, is the hidden state of the previous time node t-1, is the normalized heart rate value of the current time node t, and LSTM is an action symbol for extracting the heart rate time-dependent feature using the LSTM network.

[0185] In W411, the hidden state of the new time node is continuously calculated by Equation 6 until the hidden state of the last time node is obtained. For example, if the heart rate data is a heart rate sequence with 60 heart rates, the hidden state of the last time node is . is a 256-dimensional vector.

[0186] W412 concatenates the heart rate time-dependent feature output by the LSTM network and the normalized hearing evaluation data, as shown in Equation 7.

[0187] Equation 7.

[0188] wherein, is the feature after concatenation by W412, is the heart rate time-dependent feature output by the LSTM network. is the normalized hearing evaluation data, and n is the dimension symbol of the hearing evaluation data.

[0189] For example, if three hearing thresholds are used as the hearing evaluation data, then the hearing evaluation data is a three-dimensional vector, and Equation 7 becomes:

[0190]

[0191] Therefore, the final concatenated feature is a 259-dimensional vector.

[0192] W420 constructs a hidden layer.

[0193] Specifically, the hidden layer includes three layers. The first layer structure of the hidden layer includes 1024 neurons, and the calculation formula is Equation 8.

[0194] Equation 8.

[0195] wherein, the intermediate result output by the first layer structure, the feature after splicing by W412, the bias vector, the weight matrix of the first layer structure, the intermediate result of the structure after processing by the ReLu activation function. ReLU is the action symbol processed by the Relu activation function.

[0196] is a 1024x259 dimensional vector, is a 1024x1 dimensional bias vector, is a 259 dimensional vector, and the 1024 dimensional intermediate result is generated by matrix multiplication and then processed by the ReLu activation function to generate The role of the ReLu activation function is to zero negative values.

[0197] The second layer structure of the hidden layer includes 512 neurons, and the calculation formula is formula 9.

[0198] Formula 9.

[0199] wherein, the intermediate result output by the second layer structure, the feature after splicing, the bias vector, the weight matrix of the first layer structure, the intermediate result of the structure after processing by the ReLu activation function. ReLU is the action symbol processed by the Relu activation function.

[0200] The principle is the same as the first layer structure, which will not be repeated here.

[0201] The third layer structure of the hidden layer includes 256 neurons, and the calculation formula is formula 10.

[0202] Formula 10.

[0203] wherein, the intermediate result output by the third layer structure, the feature after splicing, the bias vector, the weight matrix of the first layer structure, the intermediate result of the structure after processing by the ReLu activation function.

[0204] ​​​​​​​​​​​​​​​​​​​​​​​​​​​The principle is the same as the first layer structure, which will not be repeated here. The hearing evaluation data directly splices the calculation of the first layer structure to ensure that the model training conforms to the specific mode of the hearing impaired population.

[0205] W420, generates a predicted HRV value according to formula 11, and the predicted HRV value is the output data of the heart rate variability prediction model.

[0206] Formula 11.

[0207] wherein, is a weight vector. is a bias scalar. is to be The structure after processing by the ReLu activation function.

[0208] For example, the input of formula 11 is is a 256-dimensional vector, and the final generated predicted HRV value is 52.3 milliseconds.

[0209] W430, calculates the loss and updates the weight-related data using the loss.

[0210] Specifically, the loss is calculated according to formula 12.

[0211] Formula 12.

[0212] wherein, is a hearing impact coefficient, which can be set to a fixed value of 0.1. is a predicted HRV value, is a standard HRV under the same heart rate data, which is a HRV value calculated by a normal traditional calculation method (without mixing hearing evaluation data). is a hearing loss degree, which is a data in the hearing evaluation data.

[0213] Of course, formula 12 is an embodiment, and formula 12 is not necessarily used to calculate the loss. In this embodiment, the more severe the hearing loss, the greater the penalty of the model to the prediction result error.

[0214] Next, the weight-related data item is adjusted according to formula 13.

[0215] Formula 13.

[0216] wherein, is the updated weight-related data, is the weight-related data before updating. is a learning rate, and the initial value of the learning rate can be set to 0.01, G is a historical gradient square cumulative amount, and W is a weight-related data item. is the loss of the partial derivative of the data item related to the weight. The data item related to the weight can be , , , .

[0217] W500, create an emotion and psychological assessment model.

[0218] W600, input the output data of the heart rate variability prediction model, the hearing assessment data, and the expression data when collecting the sample's signs data into the emotion and psychological assessment model as training data, and train the emotion and psychological assessment model; the output data of the emotion and psychological assessment model is an emotion balance index and a psychological stress index.

[0219] W600 includes:

[0220] W610, build an HRV-hearing feature channel, splice the data output by the heart rate variability prediction model and the hearing assessment data, and obtain an HRV-hearing feature. As shown in formula 14.

[0221] Formula 14.

[0222] wherein, is the feature after splicing by W610, that is, the HRV-hearing feature, is the normalized hearing assessment data, and n is the dimension symbol of the hearing assessment data.

[0223] The principle of formula 14 and formula 7 is the same, and is not repeated. If n=3, then is a four-dimensional vector.

[0224] W620, build a fully connected layer of the HRV-hearing feature according to formula 15.

[0225] Formula 15.

[0226] wherein, is the output result of the fully connected layer of the HRV-hearing feature, is the weight matrix of the fully connected layer, is the feature after splicing by W610, is the bias vector of the fully connected layer, used to adjust the baseline activation value of the fully connected layer. The fully connected layer has 128 neurons, is a 128x4 weight matrix, is a 128x1 bias vector.

[0227] W630, build an expression space-time feature channel to obtain an expression pooling feature.

[0228] W630 comprises:

[0229] W631, constructing a multi-dimensional expression feature matrix.

[0230] Specifically, in the above embodiment, the expression data comprises the mouth corner vertical offset, the pupil distance change rate and the inter-brow distance, and the time dimension of the extracted time nodes is 60, so that W631 generates an expression feature matrix of 60 time points. The mouth corner vertical offset, the pupil distance change rate and the inter-brow distance are three dimensions, so that the expression feature matrix of 60 time points is a 60x3 matrix, which has 180 elements.

[0231] W631, constructing a convolution layer, and performing 1D convolution on the multi-dimensional expression feature matrix in the convolution layer according to formula 16.

[0232] Formula 16.

[0233] wherein, is the output value of the i th time node and the j th convolution kernel. is the convolution kernel weight matrix, It can be understood that each is a vector with a length of 3. After element-by-element multiplication and summation with part of the elements in the multi-dimensional expression feature matrix output by W631, j is the number of the convolution kernel, and k is the time step offset in the convolution window, is all features of the i+k-1 th time node in the multi-dimensional expression feature matrix, that is, the mouth corner vertical offset, the pupil distance change rate and the inter-brow distance of the i+k-1 th time node. is the bias value of the j th convolution kernel.

[0234] W632, constructing a pooling layer, and performing a maximum pooling operation in the pooling layer to obtain a pooling feature according to formula 17.

[0235] Formula 17.

[0236] wherein, is the pooling feature of the i th pooling window and the j th convolution kernel, that is, the output value of the i th pooling window and the j th convolution kernel. i is the number of the pooling window, and j is the number of the convolution kernel. is the output value of the 2i-1 th time node and the j th convolution kernel, is the output value of the 2i th time node and the j th convolution kernel.

[0237] W633, performing a global average pooling operation in the pooling layer according to formula 18 to obtain a global feature.

[0238] Formula 18.

[0239] in, As a global feature, Let i be the pooling feature of the i-th pooling window and the j-th convolutional kernel, where i is the number of the pooling window. This represents the total number of pooled windows.

[0240] W640. Cross-modal fusion of HRV-hearing features and facial expression pooling features.

[0241] The W640 includes:

[0242] W641 performs cross-modal fusion according to Formula 19.

[0243] Formula 19.

[0244] in, The fusion features obtained after cross-modal fusion. The output of the fully connected layer for HRV-hearing features. This is a global feature.

[0245] W642, layer standardization is performed according to Formula 20.

[0246] Formula 20.

[0247] in, For the standardized fusion features, for The characteristic mean, for standard deviation For scaling parameters, These are translation parameters. Scaling and translation parameters are similar to learning step size or learning rate; they are learnable parameters used to adjust the normalized distribution.

[0248] W650, respectively constructs predictive formulas for the emotional balance index and the stress index.

[0249] The W650 includes:

[0250] W651, construct the prediction formula for the emotional balance index, as shown in Formula 21.

[0251]

[0252] Formula 21.

[0253] in, This is an index of emotional balance. This is the weight vector of the emotional balance index. Decision made contribution to the emotional balance index. is a bias scalar for the emotional balance index, used to adjust the baseline value of is a baseline value. Sigmoid is a sigmoid function. LayerNorm is a layer normalization function for dimension unification. is the normalized fused feature.

[0254] W652, a prediction formula for constructing the psychological stress index, as shown in equation 22.

[0255]

[0256] Equation 21.

[0257] wherein, is the emotional balance index. is a weight vector for the psychological stress index, determines contribution to the emotional balance index. is a bias scalar for the psychological stress index, used to adjust the baseline value of is a baseline value. LayerNorm is a layer normalization function for dimension unification. is the normalized fused feature.

[0258] W652, calculating the loss and updating the data related to the weights using the loss.

[0259] Specifically, this step is W430 in the heart rate variability prediction model training process, that is, the loss calculation method and weight item update of the emotional and psychological evaluation model can be designed by referring to the principles consistent with equations 12 and 13.

[0260] The application also provides a hearing aid debugging system.

[0261] As shown in Figure 2 , in an embodiment of the application, the hearing aid debugging system includes a cloud server 100 and a hearing aid 200.

[0262] The cloud server 100 is configured to execute the hearing aid debugging method mentioned in any one of the preceding embodiments. The hearing aid 200 is in communication connection with the cloud server 100.

[0263] The technical features of the above-mentioned embodiments can be combined in any manner, and the execution order of the method steps is not limited. In order to make the description simple, not all possible combinations of the technical features in the above-mentioned embodiments are described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0264] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific manner, but should not be construed as limiting the scope of the patent of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A hearing aid fitting method, characterized by, The hearing aid debugging method comprises: obtaining a reference parameter of the hearing aid and applying the reference parameter of the hearing aid to the hearing aid; increasing the hearing aid parameter by a first preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample; obtaining the emotional balance index and the psychological stress index of the sample in real time during the hearing test, and determining whether the emotional balance index and the psychological stress index increase during the hearing test; if the emotional balance index and the psychological stress index increase during the hearing test, returning to the step of increasing the hearing aid parameter by the first preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample; if the emotional balance index and the psychological stress index decrease during the hearing test, decreasing the hearing aid parameter by a second preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample; the second preset percentage is less than the first preset percentage; obtaining the emotional balance index and the psychological stress index of the sample in real time during the hearing test, and determining whether the emotional balance index and the psychological stress index increase during the hearing test; if the emotional balance index and the psychological stress index increase during the hearing test, returning to the step of decreasing the hearing aid parameter by the second preset percentage of the current value, and applying the adjusted hearing aid parameter to the hearing aid and performing a hearing test on the sample; if the emotional balance index and the psychological stress index decrease during the hearing test, outputting the current value of the hearing aid; writing the current value of the hearing aid as the optimal hearing aid parameter into the hearing aid chip of the hearing aid; after the step of outputting the current value of the hearing aid, before the step of writing the current value of the hearing aid as the optimal hearing aid parameter into the hearing aid chip of the hearing aid, the hearing aid debugging method further comprises: creating a monitoring number, and assigning an initial value of 0 to the monitoring number; taking the current value of the hearing aid as a benchmark parameter; setting a benchmark parameter fluctuation range; the benchmark parameter fluctuation range is [benchmark parameter - benchmark parameter × fourth preset percentage, benchmark parameter + benchmark parameter × fourth preset percentage]; performing a fluctuation test on the current value of the hearing aid, applying the current value of the hearing aid to the hearing aid and performing a hearing test on the sample; obtaining the emotional balance index and the psychological stress index of the sample in real time during the hearing test, and determining whether the emotional balance index and the psychological stress index of the sample during the hearing test are within the standard parameter fluctuation range; if the emotional balance index and the psychological stress index of the sample during the hearing test are within the standard parameter fluctuation range, increasing the monitoring number by 1 based on the original value, and determining whether the current value of the monitoring number is greater than or equal to a preset monitoring number; if the current value of the monitoring number is less than the preset monitoring number, after a preset time period, returning to the step of performing a fluctuation test on the current value of the hearing aid, applying the current value of the hearing aid to the hearing aid and performing a hearing test on the sample; if the value of the monitoring number is greater than or equal to the preset monitoring number, adjusting the value of the monitoring number to 0, and outputting the current value of the hearing aid; The hearing aid debugging method further comprises: W100, acquiring the physical data of the sample by using the test sound, and the expression data when the physical data is acquired; W200, calling the hearing evaluation data of the sample, comprising: W210, calling the low-frequency threshold, the medium-frequency threshold and the high-frequency threshold of the sample; W220, performing cleaning processing on the physical data of the sample, the expression data when the physical data is acquired and the hearing evaluation data; W230, performing normalization processing on the physical data after cleaning processing, the expression data after cleaning processing when the physical data is acquired and the hearing evaluation data after cleaning processing; W300, creating a heart rate variability prediction model; W400, inputting the physical data of the sample and the hearing evaluation data as training data into the heart rate variability prediction model to train the heart rate variability prediction model; W500, creating an emotion and psychological evaluation model; W600, inputting the output data of the heart rate variability prediction model, the hearing evaluation data and the expression data of the sample when the physical data is acquired as training data into the emotion and psychological evaluation model to train the emotion and psychological evaluation model; the output data of the emotion and psychological evaluation model is an emotion balance index and a psychological stress index.

2. The hearing aid fitting method according to claim 1, characterized in that, Before the reference parameter of the hearing aid is acquired and the reference parameter of the hearing aid is applied to the hearing aid, the hearing aid debugging method further comprises: determining whether the hearing aid is initially debugged; if the hearing aid is initially debugged, calling the fitter debugging parameter, and taking a preset percentage of the fitter debugging parameter as the reference parameter of the hearing aid.

3. The hearing aid fitting method according to claim 2, characterized in that, The preset percentage is 80%.

4. The hearing aid fitting method of claim 1, wherein, The output current value of the hearing aid comprises: increasing the current value of the hearing aid by a third preset percentage of the amplitude of the current value of the hearing aid, outputting the adjusted hearing aid parameter.

5. The hearing aid fitting method of claim 1, wherein, After the emotion balance index and the psychological stress index of the sample are acquired in real time during the hearing test on the sample, and it is determined whether the emotion balance index and the psychological stress index of the sample during the hearing test are within the standard parameter fluctuation range, the hearing aid debugging method further comprises: if the emotion balance index and the psychological stress index of the sample during the hearing test are not within the standard parameter fluctuation range, adjusting the number of monitoring times to 0, and returning to the increasing adjustment of the hearing aid parameter by the first preset percentage of the amplitude of the current value, which is applied to the hearing aid and used for the hearing test on the sample after adjustment.

6. The hearing aid fitting method according to claim 5, characterized in that, The hearing aid debugging method further comprises creating a cycle number, and the initial value of the cycle number is 0. After the number of monitoring times is adjusted to 0 and the current value of the hearing aid is output, the hearing aid debugging method further comprises: taking the current value of the hearing aid as a new reference parameter; increasing the cycle number by 1 based on the original value; determining whether the current value of the cycle number is greater than or equal to a preset cycle number; if the current value of the cycle number is less than the preset cycle number, returning to the acquisition of the reference parameter of the hearing aid and the application of the reference parameter of the hearing aid to the hearing aid; if the current value of the cycle number is greater than or equal to the preset cycle number, acquiring the new reference parameter obtained after each cycle. sequencing the new reference parameters obtained after each cycle in descending order, and selecting the maximum reference parameter and the minimum reference parameter; calculating the difference between the maximum reference parameter and the minimum reference parameter, and calculating the percentage of the difference in the minimum reference parameter, defining the percentage as a fluctuation percentage; determining whether the fluctuation percentage is less than or equal to a preset fluctuation percentage; if the fluctuation percentage is less than or equal to the preset fluctuation percentage, adjusting the number of cycles to 0, calculating the average of the new reference parameters obtained after each cycle, and outputting the average as the current value of the hearing aid.

7. The hearing aid fitting method according to claim 6, characterized in that, After the determination of whether the fluctuation percentage is less than or equal to the preset fluctuation percentage, the hearing aid debugging method further comprises: if the fluctuation percentage is greater than the preset fluctuation percentage, adjusting the number of cycles to 0, returning to the acquisition of the reference parameter of the hearing aid, and applying the reference parameter of the hearing aid to the hearing aid.

8. The hearing aid fitting method according to claim 7, characterized in that, After the current value of the hearing aid is written into the hearing aid chip of the hearing aid as the optimal hearing aid parameter, the hearing aid debugging method further comprises: generating a personal profile of the sample, and storing the optimal hearing aid parameter into the personal profile of the sample corresponding to the ID of the sample.

9. A hearing aid fitting system, characterized in that comprises: a cloud server configured to execute the hearing aid debugging method according to any one of claims 1 to 8; a hearing aid in communication connection with the cloud server.

Images