Moving coil vibrating diaphragm automobile audio

Abstract

The invention provides a moving coil diaphragm car audio, which is regulated and controlled in real time through safety margin detection, specifically, an alarm signal is extracted from a mixed sound field through an improved blind source separation algorithm, and sub-band decomposition covers a frequency band of 0-300Hz / 300Hz-5kHz / 5kHz-20kHz; an auditory discernible index ADI is calculated every 100 ms, and based on a Mel filter bank and an ISO 226 auditory weight, the spectrum difference between the alarm and the background at the frequency band of 1200-3000 Hz is quantized; when ADIlt; when the frequency range is 0.35, the frequency domain enhancement is started, the fundamental frequency offset is 50-100 Hz, the volume is increased by 6-12 dB, meanwhile, the voiceprint anchoring unit restrains the harmonic energy change to be smaller than or equal to 15%, the phase offset is smaller than or equal to 22.5 degrees, and identifiability is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle audio systems, and relates to a moving coil diaphragm automobile audio system, in particular to real-time regulation and control of a vehicle sound field through safety margin calculation. BACKGROUND

[0002] The moving coil diaphragm automobile audio is the most widely used type of automobile audio at present, and its core principle is that the wire wound on the voice coil skeleton moves in the magnetic field generated by the permanent magnet, drives the cone-shaped diaphragm bonded thereto to vibrate forward and backward, and thus pushes the air to produce sound. This structure is mature and reliable, the cost is controllable, and strong low-frequency effect can be achieved.

[0003] However, one potential risk of the moving coil unit is that its inherent frequency response characteristics are prone to conflict with key warning sounds of the vehicle, such as the back-up radar beep, blind spot monitoring prompt, lane departure warning sound, resulting in the driver not being able to hear or completely ignoring these vital safety warnings. The root of the problem lies in: many warning sounds are designed in the medium frequency range of 200-500Hz, and the traditional moving coil unit often has a significant resonance peak in the 100-300Hz interval in order to pursue music performance, which will greatly enhance the sound pressure level of this frequency band. When the high-decibel music signal just covers the frequency band of the warning sound, the strong music energy will form an “acoustic masking effect”, and the human ear is physiologically difficult to distinguish the sounds with similar frequencies but large differences in loudness at the same time. Even if the warning system continues to sound in the background, its sound will be drowned in the amplified music roar of the same frequency band by the moving coil unit, posing a serious safety hazard to the driver.

[0004] The existing standard stipulates that the emergency warning sound pressure level needs to reach 78±5dB(A). This standard only tests the intensity of a single warning signal in a static environment, without considering the multi-source conflict scene of music, navigation, and environmental noise in actual driving. Real vehicle test data shows that when the background music is >65dB(A), the accident response delay of the traditional scheme increases by 2.3 seconds, and the collision warning false negative rate is high in a rainstorm environment. The root cause lies in: the energy superposition model is invalid - when the music frequency spectrum energy is in the 1-3kHz interval and exceeds the warning sound, the human ear hearing masking effect leads to a decrease in the audibility of the warning.

[0005] Some existing technologies use a preset volume gain: when the warning is detected, the background music attenuation is forcibly increased by 6dB. However, this scheme causes acoustic oscillation in the tunnel low-frequency noise, resulting in a decrease in the intelligibility of voice navigation. Global volume adjustment disrupts the balance of the sound field, and cannot solve the problem of spectral overlap, therefore, it is urgent to develop an integrated system that integrates real-time conflict detection, acoustic safety evaluation, and recognizability guarantee. SUMMARY

[0006] In order to solve the above problems, the application provides a moving coil diaphragm automobile sound, which has a moving coil diaphragm loudspeaker, and realizes real-time regulation and control of the vehicle-mounted sound field through safety margin calculation, and is characterized by comprising the following modules: The real-time sound field monitoring unit: through the microphone array distributed in the cockpit, the mixed audio signal is collected and transmitted to the digital signal processor, and the signal is separated into three frequency bands: 0-300Hz, 300Hz-5kHz and 5kHz-20kHz by using sub-band decomposition technology, wherein the alarm separation adopts a blind source separation algorithm based on negative entropy maximization, and the separation matrix is optimized by iteration to ensure that the pure alarm component is extracted within 80 milliseconds.

[0007] The dynamic safety margin engine: the auditory discriminability index ADI is calculated every 100 milliseconds: first, the separated alarm and background signals are subjected to mel-scale filter bank analysis to obtain the logarithmic energy of 40 frequency bands; then, the ISO 226 standard auditory weight is applied to the 1200-3000Hz sensitive frequency band of the human ear, and the weighted spectral difference value is calculated; finally, the value is divided by the sliding average value of the background sound pressure level to generate the ADI quantitative index; Among them is the critical band weight of the ISO 226 standard weighting, and the 15th-25th channel weight coefficient, is the logarithmic energy output of the 40-channel mel filter bank, and respectively correspond to the alarm sound and the background sound, the numerator integrates the human ear sensitive frequency band 1200-3000Hz, is the Leq equivalent sound pressure level in the 200ms window of the background music; The adaptive alarm adjustment module: performs three-level response: when ADI<0.35, the frequency domain enhancement mode is started, the alarm fundamental frequency is dynamically shifted by 50-100Hz according to the dominant frequency direction of the environmental noise, and the volume is increased by 6-12dB using a feed-forward compressor; When ADI is in 0.35-0.6, only the dynamic range is compressed; When ADI>0.6, the original output is maintained.

[0008] The voiceprint anchoring unit: the harmonic phase synthesis technology is used to maintain the recognizability of the alarm, the amplitude / phase template of the original alarm is extracted, and the change of the first three harmonic energy of the adjusted signal is forced to be ≤15% and the phase shift is forced to be ≤22.5 degrees.

[0009] The signal separation adopts a fourth-order statistics joint diagonalization method, the input signal is pre-whitened to eliminate correlation, the fourth-order cumulant tensor is calculated and diagonalized through Jacobi rotation, and finally the separation matrix is derived, which is accelerated on a digital signal processor using single instruction multiple data stream, and the time consumption for processing a 128-sample-point frame is less than 15 milliseconds, and the signal-to-interference ratio of the output alarm signal is not less than 18 decibels.

[0010] The processing flow of ADI calculation: a Mel filter bank configures 40 non-uniform triangular bandpass filters, the center frequencies of which are distributed according to the human ear hearing characteristics; the band energy calculation adopts a square magnitude logarithmic conversion; the auditory weight function is generated according to the equal-loudness curve, and a double weight is given to the medium frequency band; the spectral difference integral is calculated by using the trapezoidal approximation method, and the time window length is 200 milliseconds.

[0011] Implementation of the frequency domain enhancement mode: the fractional delay filter structure is used for the fundamental frequency offset, and the filter coefficients are generated in real time through a cubic polynomial; the absolute value envelope detection is used for the dynamic range compressor, and the threshold value thereof decreases linearly with the ADI value; the gain smoothing applies a first-order recursive filter.

[0012] Workflow of the voiceprint anchoring unit: the adjusted signal is subjected to short-time Fourier transform to obtain an amplitude spectrum; the target phase spectrum is reconstructed based on the least square criterion; and the phase consistency is optimized through iterative projection.

[0013] It also includes an offline sandbox test module: 20 types of noise scene acoustic characteristics are pre-stored, including heavy rain narrowband noise, tunnel low-frequency standing wave and high-speed wide-frequency wind noise; the safety boundary is generated by using the parameter scanning method - the fixed alarm sound pressure level is 75 decibels, the background volume and the music spectral barycenter are scanned, and the critical point of ADI = 0.35 is determined by using the bisection method; the safety contour map is constructed by using the cubic spline interpolation algorithm, and is dynamically loaded to the vehicle-mounted system through the Internet of Vehicles.

[0014] A sound field safety regulation method based on the above system, comprising: (1) Signal separation stage: after the microphone signal is subjected to bandpass filtering, the source separation is realized through fourth-order cumulant diagonalization; (2) ADI calculation stage: the alarm and background signals are subjected to windowed Fourier transform, processed by a Mel filter bank, and the weighted spectral difference is calculated; (3) Dynamic adjustment stage: when ADI < 0.35, the fundamental frequency offset and the gain value are calculated, and are injected into the real-time processing pipeline; (4) Voiceprint compliance stage: the fundamental frequency offset of the output signal is verified to be ≤5% and the harmonic distortion rate is verified to be ≤3%, and the phase reconstruction is started when the limit is exceeded.

[0015] Frequency shift operation of step (3): The fractional delay is realized by using a polyphase filter structure, and filter coefficients are generated in real time by Lagrange interpolation; the delay amount is dynamically updated in proportion to the sampling rate; the processing chain includes 4 times up-sampling and anti-aliasing filtering to ensure spectral integrity.

[0016] The harmonic analysis in the voiceprint compliance stage is implemented as: The fundamental frequency detection uses autocorrelation function peak positioning; the harmonic distortion rate is calculated by the ratio of the energy of the first five harmonics; the phase consistency evaluation uses the complex exponential weighted cross-spectral analysis method.

[0017] It also includes an offline calibration method: The offline calibration process: in a standard anechoic chamber, measure the safety boundary of different music types under 20 types of noise; fit the mapping relationship between critical volume and ADI by S-shaped curve; the parameter compensation strategy is used for vehicle adaptation, and the boundary value of SUV is increased by 3 decibels, and the boundary value of convertible vehicle is decreased by 2 decibels.

[0018] The beneficial effects of the present application are: The present application fundamentally solves the safety hazard of background music and environmental noise masking played by the passive cone diaphragm loudspeaker of the emergency alarm through the dual mechanism of multi-source conflict real-time analysis and voiceprint constraint dynamic adjustment: Based on the improved blind source separation algorithm and the mel scale filter bank, the audible distinguishable index ADI model can accurately quantify the sound field safety margin of the human ear sensitive frequency band of 1200-3000Hz every 100ms period, and the risk identification sensitivity is improved compared with the traditional single signal intensity detection; The three-level adaptive adjustment strategy combined with the voiceprint anchoring technology eliminates the spectral masking while ensuring that the harmonic distortion of the alarm sound is ≤3%, so that the driver recognition accuracy is greatly improved; The safety boundary data of 20 types of noise scenes are pre-stored and parameterized compensation is performed according to the vehicle type, so as to realize the accurate adaptation of sound field regulation and vehicle cavity acoustics, and reduce the false triggering rate; finally, a whole-chain solution of "real-time monitoring-quantitative evaluation-dynamic adjustment-recognizable protection" is formed, and the collision accident rate is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0020] FIG. 1 is a schematic diagram of the overall architecture of the present application; Figure 1 FIG. 1 is a schematic diagram of the overall architecture of the present application; FIG. 1 is a schematic diagram of the overall architecture of the present application; Figure 2 FIG. 1 is a schematic diagram of the overall architecture of the present application. DETAILED DESCRIPTION Example 1: In combination Figures 1 to 2 A moving-coil diaphragm car audio has a moving-coil diaphragm loudspeaker, which is real-time regulated by a safety margin calculation for vehicle sound field, and contains the following modules: Real-time sound field monitoring unit: The mixed sound field signal is collected by a microphone array and transmitted to a DSP processing chip; this unit separates the signal into three independent frequency bands: 0-300Hz, 300Hz-5kHz, and 5kHz-20kHz through a sub-band decomposition module, wherein the FastICA algorithm is used for alarm signal separation to realize blind source separation by maximizing the negative entropy, and the iteration formula is: wherein is the whitened signal, is a nonlinear function, and the separation delay is controlled within 80ms; Dynamic safety margin engine: receives the separated alarm signal and background music signal, and calculates the auditory discriminability index ADI every 100ms: wherein is the ISO 226 standard weighted critical band weight (15-25 channel weight coefficient 2.0), is the logarithmic energy output of the 40-channel mel filter bank, and the numerator integrates the human ear sensitive frequency band 1200-3000Hz, and the denominator is the Leq equivalent sound pressure level in the 200ms window of the background music.

[0021] Adaptive alarm adjustment module: according to the ADI value, three-level operation is performed: when ADI<0.35, the frequency domain enhancement mode is activated, and the alarm fundamental frequency is changed by real-time resampling: ; wherein, the dynamic frequency shift amplitude , , adjusted alarm sound output fundamental frequency, native alarm sound design fundamental frequency, frequency shift direction decision function, real-time environmental noise dominant frequency, current alarm sound instantaneous fundamental frequency; At the same time, a dynamic range compressor is used to improve the volume: ; R is the initial compression ratio 4:1; when 0.35≤ADI≤0.6, only the compressor is enabled; ADI>0.6, the original signal is passed through; Voiceprint anchoring unit: maintain the alarm recognizable by band-limited phase synthesis technology, extract the harmonic structure template of the original alarm , adjust to force to meet: , .

[0022] Further, the signal separation module of the real-time sound field monitoring unit is implemented as: Optimize FastICA by using fourth-order cumulant joint diagonalization JADE algorithm, the specific steps include: (1) pre-whiten the input signal : , is the whitening matrix; (2) calculate the fourth-order cumulant tensor , diagonalize it by Jacobi rotation; ( (3) extract the separation matrix that makes the cumulant diagonal; (4) output independent components: ; The algorithm is accelerated by SIMD instruction on TI C66x DSP, and it takes <15ms to process 4-channel 128-point frames. The separation accuracy is measured by signal-to-interference ratio SIR: alarm signal SIR≥18dB.

[0023] Further, the ADI calculation implementation of the dynamic safety margin engine is: The Mel filter bank uses 40 triangular bandpass filters, with center frequencies distributed according to , and the energy calculation formula is: ; where is the amplitude response of the i-th filter at DFT frequency point k; the weight function is generated according to the equal response curve: ; is the sound pressure level corresponding to the i-th channel, is the filter bandwidth; the integral in the numerator is realized by the trapezoidal method: .

[0024] Further, the frequency domain enhancement mode implementation of the adaptive alarm adjustment module is: The resampling uses a Farrow structure fractional delay filter, and the transfer function is: ; where is the delay amount, For cubic polynomial coefficients, The dynamic range compressor adopts a feedforward structure: ; ; The threshold T is dynamically adjusted with ADI: Gain smoothing is achieved through a first-order IIR filter: ; Time constant , start-up time , release time .

[0025] Further, the implementation of the voiceprint anchoring unit is: Harmonic phase synthesis is completed through least square phase reconstruction: (1) The amplitude spectrum of the adjusted signal is obtained by STFT ; (2) Solve the target phase spectrum : Where is the harmonic position, , is the reference harmonic parameter; (3) Optimize the phase through the Griffin-Lim algorithm iteration; This process is activated when ADI<0.35, and the time consumption is controlled within 20ms, and the harmonic distortion THD of the output signal is guaranteed to be ≤3%.

[0026] Further, it also includes an offline sandbox test module: Pre-store 20 kinds of noise scene acoustic fingerprints, including: Heavy rain scene: noise spectrum peak 2kHz±200Hz, roll-off slope -6dB / oct; Tunnel scene: low-frequency standing wave 125Hz±20Hz, Q value≥5; High-speed wind noise: 0.5-8kHz wide frequency white noise, A-weighted sound level 68-75dB; The safety boundary generation adopts the parameter scanning method: fixing the alarm sound pressure level 75dB(A), traversing the background volume 50-90dB and the music spectrum gravity center 200-5000Hz, searching the critical point of ADI=0.35 through the dichotomy method; the contour map is obtained through the cubic spline interpolation: ; is the B-spline basis function, and the coefficient Fitted from experimental data; this data is loaded in real-time to the vehicle system via V2X communication.

[0027] Embodiment 2: A sound field safety regulation method based on the system, comprising: (1) Real-time signal separation stage: microphone array acquisition→4-32 band-pass filtering→JADE separation, including whitening, fourth-order cumulant calculation, Jacobi rotation→output independent components; (2) ADI calculation stage: 1024-point FFT on alarm / background signals respectively with Hanning window→pass through Mel filter bank, 40 channels→take logarithm to get 、 →Calculate weighted spectral difference integral; (3) Dynamic adjustment decision stage: when ADI<0.35: Frequency shift calculation: ; Resampling rate setting: ; Gain calculation: ; Generate adjustment parameters and inject into processing pipeline; (4) Voiceprint compliance stage: Harmonic analysis on output signal→compare fundamental frequency shift and harmonic distortion rate→if out of limit, start phase reconstruction.

[0028] Further, the frequency shift operation of step (3) is specifically implemented as: A polyphase fractional delay filter bank is used, and each sampling point output is calculated: Filter coefficients Generated by Lagrange interpolation: Delay amount Dynamic update: Avoid phase jump; at the same time, an anti-aliasing structure is used: upsample by 4→delay filtering→downsample by 4.

[0029] Further, the harmonic analysis of the voiceprint compliance stage is implemented as: Detect the fundamental frequency through the autocorrelation function: The first maximum position; harmonic distortion rate calculation: ; Phase consistency is evaluated by cross-spectral density: .

[0030] Further, offline calibration method is also included: 20 kinds of noise scenes are constructed in anechoic chamber, and the safety boundary of vehicle model is generated by parameter scanning: (1) The alarm sound pressure level is fixed at 75dB(A), and the background volume is changed from 50dB to 90dB with 1dB step; (2) 100 groups of ADI values are measured for each type of music (rock / classical / podcast); (3) The critical volume is fitted by Logistic regression: The parameter k represents the boundary steepness; the is pre-stored in the car system, and is used as a reference value in real-time regulation.

[0031] Thus, the above description of the embodiments is provided for the purpose of illustrating and describing it. It is not intended to exhaust or limit the disclosure. The individual elements or features of a particular embodiment are generally not limited to that particular embodiment, but are interchangeable and used in selected embodiments, if applicable, even if not specifically shown or described. In many aspects, the same elements or features can also be changed. Such changes are not considered to deviate from the disclosure, and all such modifications are intended to be included in the scope of the disclosure.

[0032] Example embodiments are provided so that the disclosure will become thorough, and will fully convey the scope to those skilled in the art. Numerous specific details are set forth such as examples of specific parts, devices, and methods, to provide a thorough understanding of embodiments of the present disclosure. Clearly, one of ordinary skill in the art will be able to practice the example embodiments without using the specific details described herein, and thus the example embodiments are not limited to a particular embodiment. In some example embodiments, well-known procedures, well-known device structures, and well-known technologies are not described in detail.

[0033] Herein, professional terms are used only for the purpose of describing specific example embodiments, and are not intended for limiting purposes. The singular forms "a" and "the" used herein can mean to include the plural forms unless the context clearly indicates the contrary. The terms "comprise" and "have" are inclusive meanings, and thus specify the existence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or combinations thereof. Unless the order of execution is explicitly indicated, the method steps, processes, and operations described herein are not interpreted as necessarily requiring the specific order discussed and shown. It should also be understood that additional or alternative steps can be employed.

Claims

1. A dynamic diaphragm car audio system, including a dynamic diaphragm speaker, which performs sound field control by dynamic safety margin calculation, characterized in that Contains the following modules: Real-time sound field monitoring unit: A microphone array distributed throughout the cockpit collects mixed audio signals and transmits them to a digital signal processor. Sub-band decomposition technology is used to separate the signals into three frequency bands: 0-300Hz, 300Hz-5kHz, and 5kHz-20kHz. Alarm separation utilizes a blind source separation algorithm based on maximizing negative entropy. By iteratively optimizing the separation matrix, the pure alarm component is extracted within 80 milliseconds. The Dynamic Safety Margin Engine calculates the Auditory Distinguishing Index (ADI) every 100 milliseconds. It first performs a Mel-scale filter bank analysis on the separated alarm and background signals to obtain the logarithmic energy of 40 frequency bands. It then applies ISO 226 standard auditory weights to the human ear-sensitive frequency range of 1200-3000 Hz and calculates the weighted spectral difference value. Finally, this value is divided by the sliding average of the background sound pressure level to generate the ADI quantitative indicator. in Critical band weights weighted for ISO 226 standard, weight coefficients for channels 15-25, is the logarithmic energy output of the 40-channel Mel filter bank, and Corresponding to the alarm sound and background sound respectively, the numerator integral covers the human ear sensitive frequency band 1200-3000Hz, the denominator Leq is the equivalent sound pressure level of background music within a 200ms window; Adaptive alarm adjustment module: performs a three-level response. When ADI < 0.35, it activates the frequency domain enhancement mode, dynamically shifting the alarm base frequency by 50-100Hz based on the dominant frequency direction of the ambient noise, and using a feedforward compressor to increase the volume by 6-12dB. When ADI is between 0.35 and 0.6, only the dynamic range is compressed; When ADI>0.6, keep the original output; Voiceprint anchoring unit: Maintains alarm recognizability through harmonic phase synthesis technology, extracts the amplitude / phase template of the native alarm, and forces the energy change of the first three harmonics of the signal after adjustment to be ≤15% and the phase shift to be ≤22.5 degrees.

2. The system according to claim 1, wherein: Signal separation uses a fourth-order statistical joint diagonalization method, and the input signal is pre-whitened to eliminate correlation; the fourth-order cumulant tensor is calculated and diagonalized through Jacobi rotation; finally, the separation matrix is ​​derived, and single-instruction multiple-data stream acceleration is used on the digital signal processor. The processing time of 128 sampling point frames is less than 15 milliseconds, and the signal-to-interference ratio of the output alarm signal is not less than 18 decibels.

3. The system according to claim 1, wherein: The ADI calculation process involves a Mel filter bank configured with 40 non-uniform triangular bandpass filters, with center frequencies distributed according to the human hearing characteristics. Band energy is calculated using a logarithmic transformation of squared magnitudes. The auditory weighting function is generated based on equal loudness curves, giving double weight to the mid-frequency band. The spectral difference integral was calculated using the trapezoidal approximation method with a time window length of 200 ms.

4. The system according to claim 1, wherein: Frequency domain enhancement mode is implemented as follows: the fundamental frequency offset uses a fractional delay filter structure, and the filter coefficients are generated in real time using a cubic polynomial. The dynamic range compressor uses absolute value envelope detection, and its threshold decreases linearly with the ADI value. Gain smoothing uses first-order recursive filtering.

5. The system according to claim 1, wherein: The workflow of the voiceprint anchoring unit: perform short-time Fourier transform on the adjusted signal to obtain the amplitude spectrum; reconstruct the target phase spectrum based on the least squares criterion; and optimize phase consistency through iterative projection.

6. The system according to claim 1, wherein: It also includes an offline sandbox testing module: it pre-stores the acoustic characteristics of 20 types of noise scenarios, including narrowband noise from heavy rain, low-frequency standing waves in tunnels, and broadband wind noise from high speeds; the safety boundary is generated using a parameter scanning method - fixing the alarm sound pressure level at 75 decibels, scanning the background volume and the center of gravity of the music spectrum, and determining the critical point of ADI=0.35 through bisection; the safety contour map is constructed using the cubic spline interpolation algorithm and dynamically loaded into the vehicle system through the Internet of Vehicles.

7. A sound field safety control method based on the system of claims 1-6, characterized in that Include: (1) Signal separation stage: After the microphone signal is bandpass filtered, source separation is achieved by diagonalization of the fourth-order cumulants; (2) ADI calculation stage: The alarm and background signals are subjected to windowed Fourier transform, processed by a Mel filter bank, and the weighted spectral difference is calculated; (3) Dynamic adjustment stage: When ADI < 0.35, the fundamental frequency offset and gain values ​​are calculated and injected into the real-time processing pipeline; (4) Voiceprint compliance stage: Verify that the fundamental frequency deviation of the output signal is ≤5% and the harmonic distortion rate is ≤3%. If the limit is exceeded, phase reconstruction is initiated.

8. The method according to claim 7, wherein Frequency shift operation in step (3): A polyphase filter structure is used to implement fractional delay, with filter coefficients generated in real time through Lagrange interpolation. The delay is dynamically updated in proportion to the sampling rate. The processing chain includes 4x upsampling and anti-aliasing filtering to ensure spectral integrity.

9. The method according to claim 7, characterized in that The harmonic analysis in the voiceprint compliance stage is implemented as follows: The fundamental frequency is detected by locating the peak of the autocorrelation function; the harmonic distortion rate is calculated by the energy ratio of the first five harmonics; and the phase consistency is evaluated by complex exponential weighted cross-spectral analysis.

10. The method according to claim 7, wherein Also included is an offline calibration method: Offline calibration process: In a standard anechoic chamber, the safety margins of different music genres under 20 types of noise are measured; the mapping relationship between critical volume and ADI is fitted using an S-shaped curve; a parameter compensation strategy is used for vehicle model adaptation, with the margin value increased by 3dB for SUV models and decreased by 2dB for convertible models.

Images