A method for evaluating the risk degree of user abnormal sound complaints

Abstract

This invention proposes a method for assessing the risk level of user complaints about abnormal noises, comprising the following steps: acquiring audio data of the abnormal noises complained about by users; calculating the energy salience A of the abnormal noise; constructing a modulation abnormal noise matrix, wherein the modulation abnormal noise matrix includes several modulation abnormal noise salience values, the value range of which is A-100% to A+100%; acquiring subjective complaint level data; correlating the modulation abnormal noise matrix with the subjective complaint level data to determine the psychological salience B; and using a neural network model to predict the risk level of abnormal noises causing user complaints in the market, wherein the risk level of user complaints is represented by the psychological salience B. This invention can predict the likelihood of users complaining about abnormal noises in the market, helping to take preventative measures and reduce the risk of after-sales costs and damage to brand image.

Description

Technical Field

[0001] This invention relates to the field of vehicle abnormal noise evaluation, and more particularly to a method for assessing the risk level of user complaints about abnormal noises. Background Technology

[0002] In the automotive industry, complaints about abnormal noises throughout vehicles have become a widespread challenge. Once a user discovers an abnormal noise, due to the masking effect, the hearing threshold for that noise is raised, making it easier for them to notice it again during subsequent use, thus intensifying their perception and frustration.

[0003] Currently, research on automotive noise issues primarily focuses on the engineering field, analyzing physical parameters obtained through sensors and testing equipment, such as sound intensity and frequency spectrum. However, these parameters may not fully reflect the user's subjective perception of the noise, failing to accurately describe the degree of annoyance experienced by users with complaining noises, or assess the level of psychological complaint. Furthermore, repeated noises can alter hearing thresholds, intensifying customer aversion and defying description using conventional physical parameters. With the advancement of psychoacoustics research, increasing attention is being paid to the impact of sound on individual psychological states and emotions, offering a new possibility for assessing user perception and the degree of complaint regarding noises.

[0004] Chinese invention CN114764526A discloses a method for evaluating the performance of vehicle abnormal noises. Under test conditions, testers subjectively evaluate the acceptability of abnormal noises inside the vehicle and record the data. Under the same test conditions as the subjective evaluation, an objective testing module detects and processes the abnormal noises inside the vehicle, obtaining sound quality parameters of the abnormal noises. The recorded subjective evaluation data is matched with the obtained sound quality parameter data to establish an evaluation matrix corresponding to the subjective evaluation data and the sound quality parameter data. This invention uses an evaluation matrix to evaluate the performance of vehicle abnormal noises, but the evaluation is overly complex and does not address the correlation between abnormal noises during the development and after-sales stages. It cannot assess the risk of abnormal noises during the development stage, which may lead to a large investment of resources to solve abnormal noise problems during development, hindering the optimization of R&D resource utilization. Summary of the Invention

[0005] The purpose of this invention is to provide a method for assessing the risk level of user complaints about abnormal noises, in order to solve the above-mentioned problems. This method can predict the likelihood of user complaints about abnormal noises in the market, which helps to take measures in advance and reduce the risk of after-sales costs and damage to brand image.

[0006] This invention proposes a method for assessing the risk level of user complaints about abnormal noises, comprising the following steps:

[0007] Step S1: Obtain the audio data of the abnormal noises reported by the user;

[0008] Step S2: Calculate the energy prominence A of the abnormal noise;

[0009] Step S3: Construct a modulation anomaly matrix, which includes several modulation anomaly salience values, the values ​​of which range from A-100% to A+100%.

[0010] Step S4: Obtain subjective complaint level data;

[0011] Step S5: Correlate the modulation anomaly matrix with the subjective complaint level data to determine the psychological salience B;

[0012] Step S6: Use a neural network model to predict the risk level of abnormal noises causing user complaints in the market. The risk level of user complaints is represented by psychological salience B.

[0013] In one embodiment, step S1 specifically includes:

[0014] Locate the abnormal noises that users complain about, identify and reproduce the abnormal noises that users complain about;

[0015] Analyze and disassemble the operating conditions that cause abnormal noises;

[0016] A data acquisition system is used to collect audio data of abnormal noises from the entire vehicle under abnormal noise conditions.

[0017] In one embodiment, when collecting abnormal noise audio data of the whole vehicle under abnormal noise conditions, the sampling rate of the data acquisition system is more than 2.56 times the analysis bandwidth.

[0018] In one embodiment, step S2 specifically includes:

[0019] The abnormal noise audio data was analyzed by FFT using signal processing software to obtain the spectrum of the abnormal noise audio, where the horizontal axis of the spectrum represents frequency.

[0020] Determine the center frequency f of the abnormal noise c The upper limit of the frequency band for complaints about abnormal noises is f. c,u And the lower limit frequency f of the frequency band for complaints about abnormal noises c,l ;

[0021] Based on the center frequency f of the abnormal sound c The upper limit of the frequency band for complaints about abnormal noises is f. c,u and / or the lower limit frequency f of the frequency band for abnormal noise complaints c,l Calculate the upper limit frequency f of the reference band u,u and the lower limit frequency of the reference band f l,l ;

[0022] According to the upper frequency limit of the frequency band for abnormal noise complaints, f c,uThe lower limit frequency of the abnormal noise complaint band is f. c,l upper limit frequency of the reference frequency band f u,u and the lower limit frequency of the reference band f l,l Calculate the energy prominence A of the abnormal noise.

[0023] In one embodiment, the signal processing software is acoustic vibration analysis software, including HEAD Artemis, LMStestlab, and / or BBM PAK.

[0024] In one embodiment,

[0025] upper limit frequency of the reference band f u,u The calculation formula is:

[0026]

[0027] Where 20Hz≤f c When ≤200Hz, C u,0 =-0.71, C i,1 =-0.40, C u,2 =2.24, C u,3 =0.017, C u,4 = -0.019; at 200Hz <d c When ≤2000Hz, C u,0 =104.27, C u,1 =3.71, C u,2 =-2.66, C u,3 = -2.41 × 10 -3 C u,4 =2.32×10 -3 When 2000Hz <f c When <10000Hz, C u,0 = -0.67, C u,1 =-0.56, C u,2 =1.59, C u,3 = -1.1 × 10 -5 C u,4 =1.01×10 -5 ;

[0028] Lower limit frequency of the reference band f l,l The calculation formula is:

[0029]

[0030] Where 20Hz≤f c When ≤200Hz, C l,0 =-11.33, C l,1 = -0.64, C l,2=1.82, C l,3 =4.11×10 -5 C l,4 = -1.04 × 10 -3 When 200Hz <f c When ≤2000Hz, C l,0 =-13.09, C l,1 =-0.77, C l,2 =1.66, C l,3 =9.6×10 -4 C l,4 =1.04×10 -3 When 2000Hz <f c When <10000Hz, C l,0 =56.60, C l,1 =1.96, C l,2 =-1.02, C l,3 = -2.89 × 10 -4 C l,4 =2.96×10 -4 ;

[0031] The formula for calculating the energy prominence A of abnormal noise is:

[0032] In one embodiment, the modulation anomaly matrix in step S3 includes eight modulation anomaly salience values, namely A-10%, A-5%, A+5%, A+10%, A+15%, A+20%, A+25%, and A+30%.

[0033] In one embodiment, in step S4, the tester evaluates the degree of subjective complaint on the abnormal noise audio after multiple playbacks to obtain subjective complaint level data, specifically including:

[0034] Test participants were selected based on gender, age, education level, and experience in acoustic evaluation. Ages were divided into 20-29 years old, 30-39 years old, 40-49 years old, and 50 years old and above. Education levels were divided into associate degree, bachelor's degree, master's degree, and doctoral degree.

[0035] After playing back the abnormal noise audio more than 20 times, the testers were evaluated on their subjective level of complaint. The subjective level of complaint included 10 scores, namely 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10. The lower the score, the higher the testers' level of complaint about the abnormal noise. 1-5 indicates that the testers found the abnormal noise unacceptable, and 6-10 indicates that the testers found the abnormal noise acceptable.

[0036] In one embodiment, step S5 specifically includes:

[0037] Modulate the abnormal noise audio so that the modulated abnormal noise audio satisfies the modulation abnormal noise prominence;

[0038] The testers subjectively evaluated the degree of complaint regarding the modulated abnormal noise audio;

[0039] The modulated abnormal noise audio that is closest to the subjective complaint level of the abnormal noise audio after multiple playbacks is selected, and its corresponding modulated abnormal noise prominence A+x% is the psychological prominence of the abnormal noise B.

[0040] In one embodiment, the neural network model in step S6 is a BP neural network;

[0041] The input variables of the neural network model include the abnormal noise center frequency, bandwidth, energy prominence, and / or the degree of subjective complaint;

[0042] The output variables of the neural network model include psychological salience.

[0043] Compared with the prior art, the beneficial effects of the risk assessment method for user noise complaints of the present invention are as follows:

[0044] 1) This invention proposes a new psychoacoustic evaluation index, which uses psychological salience to describe the degree of users' psychological perception of abnormal noises. This can more accurately assess the degree of user complaints, establish a connection between objective data and users' subjective psychological perceptions, solve the problem of large discrepancies between users' subjective perceptions and objective data, guide enterprises to qualitatively measure the degree of customer complaints about abnormal noises, and help R&D personnel to more comprehensively understand the actual impact of abnormal noises on user experience, thereby making targeted improvements and optimizations.

[0045] 2) This invention uses neural networks to establish a connection between abnormal noise and psychological salience. For abnormal noises discovered in the early stages of R&D, it assesses the risk of customer complaints caused by such noises after the product is launched, thereby guiding the allocation of appropriate resources to analyze and resolve the issues. This reduces customer complaints caused by abnormal noises, identifies and resolves potential acoustic problems earlier, reduces the risk of after-sales costs and brand image damage, improves the utilization rate of R&D resources, optimizes R&D expenses, and reduces unnecessary resource waste. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating a method for assessing the risk level of user noise complaints according to an embodiment of the present invention.

[0047] Figure 2 This is a spectrum diagram of abnormal noise according to an embodiment of the present invention;

[0048] Figure 3 This is a structural diagram of the neural network model in a user noise complaint risk assessment method according to an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the specific embodiments described herein are for the purpose of aiding understanding the invention and do not constitute a limitation thereof.

[0050] This invention proposes a method for assessing the risk level of user complaints about abnormal noises, see [link to relevant documentation]. Figure 1 It includes the following steps:

[0051] Step S1: Obtain the audio data of the abnormal noises reported by the user;

[0052] Step S2: Calculate the energy prominence A of the abnormal noise;

[0053] Step S3: Construct a modulation anomaly matrix, which includes several modulation anomaly salience values, the values ​​of which range from A-100% to A+100%.

[0054] Step S4: Obtain subjective complaint level data;

[0055] Step S5: Correlate the modulation anomaly matrix with the subjective complaint level data to determine the psychological salience B;

[0056] Step S6: Use a neural network model to predict the risk level of abnormal noises causing user complaints in the market. The risk level of user complaints is represented by psychological salience B.

[0057] It should be noted that in the fields of acoustics and perception, salience is used to describe the degree of salience or intensity of a sound in perception.

[0058] The following section elaborates on the risk assessment methods for user complaints about abnormal noises.

[0059] Step S1 of one embodiment of the present invention specifically includes:

[0060] To pinpoint the unusual noises complained about by users, and given the large number of such complaints in the market, we will use methods such as telephone follow-ups and on-site visits to understand the specifics of the unusual noises, and reproduce and identify the abnormal noises complained about by users.

[0061] The abnormal noises complained about by users and their corresponding operating conditions (such as specific vehicle speeds, engine loads, etc.) are broken down and simplified into abnormal noise conditions that can be repeatedly achieved in engineering.

[0062] A data acquisition system is used to collect audio data of abnormal noises from the entire vehicle under abnormal noise conditions: the sound pressure signal of the abnormal noise is converted into an analog electrical signal using a microphone and recorded by the data acquisition system. To ensure high-quality data acquisition, the sampling rate of the data acquisition system should be set to at least 2.56 times the analysis bandwidth to prevent aliasing (aliasing refers to signal distortion caused when the signal frequency is higher than half of the sampling rate).

[0063] Step S2 of one embodiment of the present invention specifically includes:

[0064] The abnormal noise audio data was analyzed using FFT software to obtain the spectrum of the abnormal noise, such as... Figure 2 As shown, the horizontal axis of the spectrum represents frequency, and the vertical axis represents the signal strength or amplitude at that frequency;

[0065] The abnormal noise was reproduced using an audio playback device, and the center frequency f of the noise was determined by performing spectrum analysis on the noise data. c The upper limit of the frequency band for complaints about abnormal noises is f. c,u And the lower limit frequency f of the frequency band for complaints about abnormal noises c,l ;

[0066] Based on the center frequency f of the abnormal sound c The upper limit of the frequency band for complaints about abnormal noises is f. c,u and / or the lower limit frequency f of the frequency band for abnormal noise complaints c,l Calculate the upper limit frequency f of the reference band u,u and the lower limit frequency of the reference band f l,l ;

[0067] According to the upper frequency limit of the frequency band for abnormal noise complaints, f c,u The lower limit frequency of the abnormal noise complaint band is f. c,l upper limit frequency of the reference frequency band f u,u and the lower limit frequency of the reference band f l,l Calculate the energy prominence A of the abnormal noise.

[0068] The Fast Fourier Transform (FFT) is an algorithm used to calculate the Fourier transform, converting a time-domain signal into a frequency-domain representation through a series of mathematical operations. The Fourier transform is a method for converting a signal from the time domain to the frequency domain, allowing us to analyze the different frequency components contained in the signal.

[0069] In one embodiment of the present invention, the signal processing software is a mature acoustic vibration analysis software, such as HEADArtemis, LMS testlab, and / or BBM PAK, etc.

[0070] The upper limit frequency f of the reference frequency band in one embodiment of the present invention u,u The calculation formula is:

[0071]

[0072] Where 20Hz≤f c When ≤200Hz, C u,0 =-0.71, C u,1 =-0.40, C u,2 =2.24, C u,3 =0.017, C u,4 = -0.019; at 200Hz <f c When ≤2000Hz, C u,0 =104.27, C u,1 =3.71, C u,2 =-2.66, C u,3 = -2.41 × 10 -3 C u,4 =2.32×10 -3 When 2000Hz <f c When <10000Hz, C u,0 = -0.67, C u,1 =-0.56, C u,2 =1.59, C u,3 = -1.1 × 10 -5 C u,4 =1.01×10 -5 .

[0073] Lower limit frequency of the reference band f l,l The calculation formula is:

[0074]

[0075] Where 20Hz≤f c When ≤200Hz, C l,0 =-11.33, C l,1 = -0.64, C l,2 =1.82, C l,3 =4.11×10 -5 C l,4 = -1.04 × 10 -3 When 200Hz <f c When ≤2000Hz, C l,0 =-13.09, C l,1 =-0.77, C l,2 =1.66, C l,3 =9.6×10 -4 C l,4 =1.04×10 -3 When 2000Hz <f cWhen <10000Hz, C l,0 =56.60, C l,1 =1.96, C l,2 =-1.02, C l,3 = -2.89 × 10 -4 C l,4 =2.96×10 -4 .

[0076] The formula for calculating the energy prominence A of abnormal noise is:

[0077] In step S3 of one embodiment of the present invention, the modulation anomaly matrix consists of eight modulation anomaly saliences, namely A-10%, A-5%, A+5%, A+10%, A+15%, A+20%, A+25%, and A+30%.

[0078] Step S4 of one embodiment of the present invention involves obtaining subjective complaint level data by having test personnel subjectively evaluate the degree of complaint based on the abnormal noise audio after multiple playbacks. Specifically, this includes:

[0079] Test subjects were selected based on gender (male / female), age (20-29, 30-39, 40-49, 50 and above), education level (associate degree, bachelor's degree, master's degree, doctoral degree, other), and whether they had experience in acoustic evaluation.

[0080] After playing the abnormal noise audio more than 20 times consecutively, the participants were assessed on their subjective level of complaint, and a subjective score was given based on Table 1. Multiple playbacks were used to reinforce the sound and simulate the psychological state of users after repeatedly hearing the abnormal noise. The subjective score consisted of 10 points: 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10. Lower scores indicated a higher level of complaint. Points 1-5 represented the unacceptable range for the abnormal noise, while points 6-10 represented the acceptable range.

[0081] Table 1 Subjective Evaluation Table

[0082]

[0083] Step S5 of one embodiment of the present invention specifically includes:

[0084] The abnormal noise audio is modulated using audio post-processing software or Matlab, and the amplitude of the abnormal noise complaint frequency band is increased or decreased after filtering, so that the modulated abnormal noise audio meets the modulation abnormal noise prominence. This is to add gain to the abnormal noise complaint frequency band by modulating the abnormal noise audio, so that the abnormal noise complaint frequency band presents different prominence and provides different auditory performances for the testers. The unmodulated abnormal noise audio is referred to as the original complaint abnormal noise in the following text.

[0085] The testers subjectively evaluated the degree of complaint regarding the modulated abnormal noise audio;

[0086] The modulated audio of the abnormal noise that most closely matches the subjective level of complaint after repeated playback of the original abnormal noise is selected. The corresponding modulated abnormal noise salience A+x% is the psychological salience B of the abnormal noise. This is because after evaluating the original abnormal noise, the test subjects' sensitivity to the abnormal noise increases due to repeated identification of the same abnormal noise, which intensifies their psychological aversion. The psychological salience B can well represent the degree of psychological gain from repeatedly hearing the abnormal noise.

[0087] In one embodiment of the present invention, the neural network model in step S6 is a BP neural network. The input variables of the neural network model include the abnormal center frequency, bandwidth, energy salience, and / or subjective rating, etc., and the output variable includes psychological salience. Its structural form is described in [reference needed]. Figure 3 Cross-iterative validation is performed between each hidden layer node and each parameter. The BP (Backpropagation) neural network is one of the most common neural networks; it is a supervised learning neural network model typically used for regression and classification tasks.

[0088] Before the neural network model is formally used to evaluate abnormal noise audio, it needs to be trained. This mainly includes: collecting abnormal noise audio that causes user complaints; calculating parameters such as the energy salience, center frequency, and bandwidth of the abnormal noise audio; modulating the abnormal noise audio to obtain a modulation abnormal noise matrix; having testers subjectively rate the degree of complaint for the abnormal noise audio and the modulated audio; and calculating the psychological salience of the abnormal noise audio. After collecting a number of the above training samples, the model is trained with 2000 training iterations and a training objective of 0.01 until the model converges.

[0089] It should be noted that the terms “comprising,” “including,” and “containing” used in this application indicate the presence of the claimed feature, but do not exclude the presence of one or more other features; the term “and / or” includes any and all combinations of one or more of the relevant listed items.

[0090] The present invention has the following beneficial effects:

[0091] 1) This invention proposes a new psychoacoustic evaluation index, which uses psychological salience to describe the degree of users' psychological perception of abnormal noises. This can more accurately assess the degree of user complaints, establish a connection between objective data and users' subjective psychological perceptions, solve the problem of large discrepancies between users' subjective perceptions and objective data, guide enterprises to qualitatively measure the degree of customer complaints about abnormal noises, and help R&D personnel to more comprehensively understand the actual impact of abnormal noises on user experience, thereby making targeted improvements and optimizations.

[0092] 2) This invention uses neural networks to establish a connection between abnormal noise and psychological salience. For abnormal noises discovered in the early stages of R&D, it assesses the risk of customer complaints caused by such noises after the product is launched, thereby guiding the allocation of appropriate resources to analyze and resolve the issues. This reduces customer complaints caused by abnormal noises, identifies and resolves potential acoustic problems earlier, reduces the risk of after-sales costs and brand image damage, improves the utilization rate of R&D resources, optimizes R&D expenses, and reduces unnecessary resource waste.

[0093] The embodiments described above are merely further illustrations of the present invention and are not intended to limit the present invention in any other way. The present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding modifications and changes based on the present invention, but all such modifications and changes should fall within the protection scope of the present invention.

Claims

1. A method for assessing the risk level of user complaints about abnormal noises, characterized in that, Includes the following steps: Step S1: Obtain the audio data of the abnormal noises reported by the user; Step S2: Calculate the energy prominence A of the abnormal noise; Step S3: Construct a modulation anomaly matrix, which includes several modulation anomaly salience values, the values ​​of which range from A-100% to A+100%. Step S4: Obtain subjective complaint level data; Step S5: Correlate the modulation anomaly matrix with the subjective complaint level data to determine the psychological salience B; Step S6: Use a neural network model to predict the risk level of abnormal noises causing user complaints in the market. The risk level of user complaints is represented by psychological salience B. Step S1 specifically includes: Locate the abnormal noises that users complain about, identify and reproduce the abnormal noises that users complain about; Analyze and disassemble the operating conditions that cause abnormal noises; Use a data acquisition system to collect audio data of abnormal noises from the entire vehicle under abnormal noise conditions; Step S2 specifically includes: The abnormal noise audio data was analyzed by FFT using signal processing software to obtain the spectrum of the abnormal noise audio, where the horizontal axis of the spectrum represents frequency. Determine the center frequency f of the abnormal noise c The upper limit of the frequency band for complaints about abnormal noises is f. c,u And the lower limit frequency f of the frequency band for complaints about abnormal noises c,l ; Based on the center frequency f of the abnormal sound c The upper limit of the frequency band for complaints about abnormal noises is f. c,u and / or the lower limit frequency f of the frequency band for abnormal noise complaints c,l Calculate the upper limit frequency f of the reference band u,u and the lower limit frequency of the reference band f l,l ; According to the upper frequency limit of the frequency band for abnormal noise complaints, f c,u The lower limit frequency of the abnormal noise complaint band is f. c,l upper limit frequency of the reference frequency band f u,u and the lower limit frequency of the reference band f l,l Calculate the energy prominence A of the abnormal noise; upper limit frequency of the reference band f u,u The calculation formula is: ； Among them, when hour, , , , , ;when hour, , , , , ;when hour, , , , , ; Lower limit frequency of the reference band f l,l The calculation formula is: ； Among them, when hour, , , , , ;when hour, , , , , ;when hour, , , , , ; The formula for calculating the energy prominence A of abnormal noise is: .

2. The method for assessing the risk level of user abnormal noise complaints according to claim 1, characterized in that, When collecting abnormal noise audio data of the whole vehicle under abnormal noise conditions, the sampling rate of the data acquisition system is more than 2.56 times the analysis bandwidth.

3. The method for assessing the risk level of user abnormal noise complaints according to claim 1, characterized in that, The signal processing software is acoustic vibration analysis software, including HEAD Artemis, LMS testlab, and / or BBM PAK.

4. The method for assessing the risk level of user abnormal noise complaints according to claim 1, characterized in that, In step S3, the modulation anomaly matrix includes eight modulation anomaly prominence levels, namely A-10%, A-5%, A+5%, A+10%, A+15%, A+20%, A+25%, and A+30%.

5. The method for assessing the risk level of user abnormal noise complaints according to claim 1, characterized in that, In step S4, the testers conduct a subjective complaint assessment of the abnormal noise audio after multiple playbacks to obtain subjective complaint level data, specifically including: Test participants were selected based on gender, age, education level, and experience in acoustic evaluation. Ages were divided into 20-29 years old, 30-39 years old, 40-49 years old, and 50 years old and above. Education levels were divided into associate degree, bachelor's degree, master's degree, and doctoral degree. After playing back the abnormal noise audio more than 20 times, the testers were evaluated on their subjective level of complaint. The subjective level of complaint included 10 scores, namely 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10. The lower the score, the higher the testers' level of complaint about the abnormal noise. 1-5 indicates that the testers found the abnormal noise unacceptable, and 6-10 indicates that the testers found the abnormal noise acceptable.

6. The method for assessing the risk level of user abnormal noise complaints according to claim 5, characterized in that, Step S5 specifically includes: Modulate the abnormal noise audio so that the modulated abnormal noise audio satisfies the modulation abnormal noise prominence; The testers subjectively evaluated the degree of complaint regarding the modulated abnormal noise audio; The modulated abnormal noise audio that is closest to the subjective complaint level of the abnormal noise audio after multiple playbacks is selected, and its corresponding modulated abnormal noise prominence A+x% is the psychological prominence of the abnormal noise B.

7. The method for assessing the risk level of user abnormal noise complaints according to claim 1, characterized in that, The neural network model in step S6 is a BP neural network; The input variables of the neural network model include the abnormal noise center frequency, bandwidth, energy prominence, and / or the degree of subjective complaint; The output variables of the neural network model include psychological salience.

Images