Method and device for optimizing sound effect in passenger car cabin

Abstract

The invention discloses a method and device for optimizing the sound effect in a passenger vehicle cabin, and relates to the technical field of vehicle-mounted audio signal processing, and the method comprises the steps: obtaining an initial reference response vector representing the factory ideal state of a loudspeaker; when the monitoring trigger condition is met, controlling a power amplifier to generate a standard test excitation signal to drive a loudspeaker group, and constructing a current state response vector through a voice acquisition unit; constructing a spectrum deviation vector based on the difference value, and eliminating the frequency response interference of the acquisition unit by using a relative measurement mechanism; calling a preset sound field transfer function matrix model, taking the spectrum deviation vector as input, and calculating a channel gain adjustment vector through an inverse problem solving algorithm containing regularization constraint; and finally, updating audio processing parameters according to the vector and executing closed-loop checking. According to the invention, by using the existing vehicle-mounted hardware, automatic monitoring and accurate compensation of the aging characteristic of the loudspeaker are realized, and the sound effect consistency in the vehicle life cycle is effectively maintained.

Description

A method and apparatus for optimizing the sound effects in the cabin of a passenger vehicle. Technical Field

[0001] This application relates to the field of vehicle audio signal processing technology, and in particular to a method and apparatus for optimizing the sound effects in the cabin of a passenger vehicle. Background Technology

[0002] With the continuous improvement of automotive intelligence and comfort, in-vehicle audio-visual entertainment systems have become an important indicator of the quality of passenger vehicle cabins. Modern high-end in-vehicle audio systems are typically equipped with multi-channel power amplifiers and speaker arrays distributed in different locations within the cabin, aiming to provide users with a high-fidelity listening experience through precise crossover and sound field tuning. During the vehicle manufacturing process, engineers meticulously calibrate the frequency response curve, phase, and gain of the audio processor to suit the specific acoustic environment of the cabin, ensuring the audio system operates at its optimal state.

[0003] However, as an electroacoustic transducer, the physical characteristics of a loudspeaker are not static throughout the long lifespan of a vehicle. Due to drastic changes in temperature and humidity within the vehicle cabin, vibrations, and mechanical fatigue from prolonged operation, the loudspeaker's suspension system (such as the surround and centering bracket) gradually ages, leading to reduced compliance or altered damping characteristics. This physical aging directly manifests as a drift in the loudspeaker's frequency response, particularly a decrease in low-frequency sensitivity or a shift in resonant frequency. This causes the actual listening experience to gradually deviate from the factory settings, resulting in muddy bass, a collapsed soundstage, or an imbalance between the three frequency ranges.

[0004] Current conventional solutions to this problem primarily rely on manual adjustments based on the user's auditory perception. Users typically need to access the settings interface of the in-vehicle entertainment system and make rough gain adjustments to the equalizer (EQ) based on their subjective feelings. However, this method not only requires users to have a certain level of acoustic appreciation, but manual adjustments can usually only perform global gain adjustments across a wide frequency range (such as bass, midrange, and treble), failing to provide precise compensation for the physical attenuation at specific frequency points. Furthermore, the in-vehicle sound field is a complex multi-input multi-output system; abnormal sound quality in a single frequency band is often the result of coupling effects from multiple physical channels. Manual adjustments not only fail to restore balance but may also lead to phase distortion or further deterioration of the sound quality due to improper parameter settings.

[0005] While some high-end models have begun to introduce active sound field calibration technology based on microphone feedback, existing technologies typically require the use of high-precision measurement-grade microphones to acquire absolute sound pressure level (SPL) data. However, due to cost control and integration considerations, microphones in ordinary passenger vehicles primarily serve voice calls or voice control functions, and their frequency response curves are often not flat, exhibiting consistency deviations under high and low temperature environments. If such non-measurement-grade microphones are directly used for absolute SPL monitoring, the system cannot distinguish whether the deviation in the measurement data stems from speaker aging or microphone error, easily leading to incorrect compensation decisions. Furthermore, existing automatic adjustment schemes lack mathematical decoupling of the complex mapping relationship between acoustic deviations and physical control channels, making it difficult to achieve precise parameter correction for specific physical channels while ensuring system stability. Therefore, how to achieve high-precision sound aging monitoring and automated closed-loop compensation using existing automotive resources without adding expensive additional sensor hardware has become a pressing technical challenge in the field of automotive audio. Summary of the Invention

[0006] The purpose of this invention is to provide a method and apparatus for optimizing the sound effects in the cabin of a passenger vehicle, at least to solve the technical problems of frequency response drift and sound quality degradation caused by physical aging of speaker components in existing in-vehicle audio systems during long-term use, which are difficult to accurately restore by manual adjustment by the user, and existing automatic calibration schemes are costly or cannot eliminate the frequency response error interference of ordinary in-vehicle microphones themselves.

[0007] This invention provides the following solution:

[0008] The first aspect of this invention provides a method for optimizing the sound effects inside a passenger vehicle cabin. The method mainly includes the following steps:

[0009] First, the initial reference response vector stored in the non-volatile memory module is acquired. This initial reference response vector is reference data established at the vehicle's factory or standard condition, characterizing the ideal frequency domain acoustic features received at the voice acquisition unit when the speaker assembly is driven by a standard test excitation signal in an unaged state. This vector serves as the physical anchor point for system calibration, recording the overall system response, which includes the acoustic characteristics of the vehicle cabin environment and the circuit transmission characteristics.

[0010] During the vehicle's service life, when the monitoring trigger conditions are met and the cabin acoustic environment meets the access standards, the system enters monitoring mode. At this time, the control power amplifier directly generates the standard test excitation signal to drive the speaker assembly to emit sound, and the onboard voice acquisition unit collects real-time acoustic signals from the cabin. The system performs frequency domain analysis on the collected signals to construct the current state response vector. During this process, to eliminate the interference of the onboard voice acquisition unit's own non-flat frequency response characteristics on the measurement results, the excitation signal type, signal source path, and gain calibration value when establishing the current state response vector must be strictly consistent with those when establishing the initial reference response vector.

[0011] Next, a spectral deviation vector is constructed based on the difference between the initial reference response vector and the current state response vector. This step employs relative measurement logic, which calculates the change in sound pressure level at the same frequency point across different time dimensions, ensuring that the inherent sensitivity error of the speech acquisition unit and the constant component in the environmental transfer function cancel each other out during the differential operation. The resulting spectral deviation vector can thus simply characterize the drift in the physical performance of the speaker assembly.

[0012] Subsequently, a preset sound field transfer function matrix model is invoked. This matrix model describes the linear mapping relationship between the gain changes of each independent control channel in the speaker array and the sound pressure level at each monitoring center frequency point, and its numerical distribution contains the crossover design characteristics of the vehicle audio system. Using the constructed spectral deviation vector as the input observation, the channel gain adjustment vector is calculated through an inverse problem-solving algorithm. This calculation process employs the least squares method with regularization constraints to minimize the frequency response deviation energy while limiting the gain adjustment amplitude, thereby solving for the target gain compensation value required for each physical channel.

[0013] Finally, based on the calculated channel gain adjustment vector, the audio processing parameters of the power amplifier are updated, and targeted gain compensation is performed on the drive signal of the speaker group to counteract the acoustic attenuation caused by physical aging.

[0014] Furthermore, during the establishment and real-time monitoring of the initial reference response vector, pink noise is used as the standard test excitation signal. Utilizing its power spectral density's inverse relationship with frequency, a balanced energy distribution is achieved on the logarithmic frequency coordinate axis. Simultaneously, the system implements strict audio channel control, forcibly cutting off external entertainment audio sources and bypassing all user-defined sound effect settings (such as EQ equalizers and surround sound settings), establishing a direct link from the digital signal generator to the digital-to-analog converter to ensure the physical purity of the measurement data.

[0015] Furthermore, a dead-zone threshold determination and weighted preprocessing mechanism are introduced when constructing the spectral deviation vector. For frequency points with a variation amplitude less than the preset noise tolerance threshold, their minor fluctuations are ignored; for low-frequency and high-frequency bands that are sensitive to speaker aging, higher weighting coefficients are assigned. This ensures that the optimization algorithm prioritizes correcting frequency bands that have a significant impact on listening experience or suffer from severe physical attenuation, avoiding system parameter oscillations caused by measurement random noise.

[0016] Furthermore, the inverse problem-solving algorithm finds the optimal solution by constructing an objective function. This objective function includes a squared Euclidean distance term (representing the residual) between the actual compensation effect and the target deviation, and a Tikhonov regularization term used to constrain the norm of the solution. By adjusting the regularization parameter, a balance is achieved between reducing acoustic errors and maintaining system stability, preventing excessive gain adjustment that could lead to speaker overload or clipping. Simultaneously, asymmetric safety boundary constraints are applied to the calculated gain adjustment values ​​to ensure that the final output parameters are within the hardware-safe operating range.

[0017] Furthermore, the method also includes a closed-loop verification mechanism. After applying the gain adjustment parameters, the system executes the excitation and acquisition process again to calculate the residual deviation after verification. Only when the residual deviation converges to within a preset threshold range are the parameters fixed to non-volatile memory; if convergence is not achieved, a limited number of iterations for fine-tuning are performed; if convergence still fails after multiple iterations, a circuit breaker mechanism is triggered and the parameters are rolled back to prevent algorithm divergence due to hardware failure.

[0018] A second aspect of the present invention provides an apparatus for optimizing the sound effects inside a passenger vehicle cabin. The device includes: a storage module for storing the initial reference response vector, the sound field transfer function matrix model, and preset control thresholds and calibration parameters; an environmental monitoring and access module for acquiring vehicle speed, gear position, window status, and in-vehicle background noise data, determining whether the current environment meets the acoustic testing access standards, and responsible for switching and bypassing audio signal paths; a signal processing and analysis module configured to generate standard test excitation signals to drive the speaker assembly, and perform analog-to-digital conversion, windowing, framing, and spectral analysis on the analog signals acquired by the voice acquisition unit to construct the current state response vector, and further calculate the spectral deviation vector relative to the reference state; a calculation and optimization module for running an inverse problem solving algorithm, based on the input spectral deviation vector and the pre-stored sound field transfer function matrix model, calculating the channel gain adjustment vector corresponding to each physical channel by minimizing the objective function containing regularization terms; and a closed-loop control module for performing smooth updates of gain parameters, triggering secondary verification after parameter adjustment, and determining whether to perform parameter solidification storage, iterative optimization, or fault rollback operation based on the residual convergence.

[0019] The above solution achieves the following beneficial technical effects:

[0020] This invention employs a spectrum deviation construction method based on the principle of relative measurement. By calculating the sound pressure level difference at the same frequency point in the initial reference state and the current aging state, the inherent non-flat frequency response characteristics, sensitivity errors, and transfer function components of the vehicle cabin's static acoustic environment of the in-vehicle voice acquisition unit are mutually canceled out in the differential operation. This technical feature avoids dependence on expensive high-precision measurement-grade microphones, and can accurately capture the physical performance drift of speaker components caused by aging using the vehicle's existing ordinary voice communication microphones, thus reducing the system's hardware cost and implementation threshold.

[0021] This invention utilizes a pre-defined acoustic field transfer function matrix model combined with a least-squares algorithm with regularization constraints to solve the inverse problem, accurately mapping the high-dimensional acoustic spectrum deviation vector to the electrical gain adjustment vectors of each independent channel of the loudspeaker. This method leverages the physical frequency division characteristics of the transfer matrix to achieve automatic decoupling of frequency components from physical control channels. Simultaneously, the Tikhonov regularization term and dead-zone threshold determination logic effectively suppress the interference of measurement noise on the calculation results, preventing system parameter oscillations caused by overfitting and ensuring the mathematical convergence and physical feasibility of the multi-channel gain compensation scheme.

[0022] This invention integrates parameter boundary constraints, smooth gain transition, and closed-loop residual verification mechanisms. After calculating the theoretical gain, it enforces amplitude limits and performs secondary sound field verification after parameter updates. This closed-loop control logic not only effectively prevents the risk of speaker overload, clipping, or voice coil burnout caused by aggressive compensation, but also eliminates abrupt noise during the adjustment process through smooth switching. This ensures that while automatically repairing sound quality aging, it also maintains the hardware safety of the in-vehicle audio system and the user's actual listening experience. Attached Figure Description

[0023] Figure 1 is a schematic diagram of the hardware architecture of a method and apparatus for optimizing the sound effects in the cabin of a passenger vehicle, provided by one or more embodiments of the present invention.

[0024] Figure 2 is a general flowchart of a method and apparatus for optimizing the sound effects in the cabin of a passenger vehicle provided by one or more embodiments of the present invention.

[0025] Figure 3 is a closed-loop optimization logic block diagram based on spectrum deviation and transfer matrix of a method and apparatus for optimizing the cabin sound effect of a passenger car provided by one or more embodiments of the present invention. Detailed Implementation

[0026] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Referring to Figure 1, the present invention provides a passenger vehicle cabin audio optimization system 100, which includes: a cabin domain controller 110, a digital audio bus 120, a power amplifier 130, a voice acquisition unit 140, and a speaker assembly 150. The cabin domain controller 110, the power amplifier 130, and the voice acquisition unit 140 are physically connected via the digital audio bus 120. The digital audio bus 120 adopts a daisy-chain topology for synchronous transmission of high-bandwidth digital audio data and control commands between connected nodes. In one embodiment, the digital audio bus 120 is an A2B bus.

[0028] In the communication architecture of the digital audio bus 120, the cockpit domain controller 110 is configured as the master node, responsible for managing the vehicle's audio strategy, displaying the human-machine interface, and generating the system clock. The power amplifier 130 and the voice acquisition unit 140 are configured as slave nodes, responsible for executing specific audio driving, signal amplification, and acoustic signal acquisition tasks.

[0029] The voice acquisition unit 140 is located at a predetermined position inside the vehicle cabin. Specifically, the voice acquisition unit 140 is an in-vehicle voice microphone used to pick up ambient sounds and test excitation signals within the vehicle cabin. The voice acquisition unit 140 is configured to convert the acquired analog acoustic signals into a digital audio stream and load it into the uplink time slot of the digital audio bus 120 for transmission. The power amplifier 130 is connected between the digital audio bus 120 and the speaker assembly 150. The power amplifier 130 integrates a digital signal processing module 131 (DSP) and a non-volatile memory module 132.

[0030] The digital signal processing module 131 is configured to execute the core algorithm for sound effect aging compensation. Specifically, the digital signal processing module 131 directly acquires the real-time digital audio stream acquired by the voice acquisition unit 140 from the digital audio bus 120 via bus data listening or data loopback. The non-volatile storage module 132 is used to store initial reference data. This initial reference data includes the reference response vector measured during the vehicle's manufacturing or calibration phase. .vector Defined as:

[0031] ;

[0032] in, The total number of monitoring frequency points, ( ) indicates the first The sound pressure level values ​​at each center frequency point under reference conditions.

[0033] The power amplifier 130 is also configured to adjust the drive signal parameters output to the speaker group 150 based on the calculation results of the digital signal processing module 131.

[0034] The speaker assembly 150 consists of multiple speaker units arranged in different spatial locations within the vehicle cabin. The speaker assembly 150 is divided into... Each of the following is a separate control channel, and each channel is connected to an output port of the power amplifier 130. The control channels include, but are not limited to, the left front door bass channel, the right front door bass channel, the left rear door tweeter channel, the right rear door tweeter channel, and the subwoofer channel.

[0035] The system 100 also includes a human-machine interface module 160, which is communicatively connected to the cockpit domain controller 110. The human-machine interface module 160 is configured to receive audio calibration commands from users and send monitoring trigger signals to the power amplifier 130 via the cockpit domain controller 110.

[0036] During the operation of system 100, the power amplifier 130 acts as an excitation source, sending a preset standard test signal to the speaker assembly 150. The voice acquisition unit 140 simultaneously acquires the sound field response signal within the vehicle cabin. The digital signal processing module 131 compares the real-time acquired sound field response signal with the reference response vector stored in the non-volatile storage module 132. The spectral deviation is calculated, and the gain parameters of each speaker channel are adjusted based on the deviation.

[0037] Through the aforementioned hardware architecture, this invention utilizes the vehicle's existing voice acquisition and audio transmission links to construct a closed-loop control system from excitation playback, signal acquisition, data processing to parameter adjustment, thereby realizing digital monitoring and compensation for the physical performance degradation of the speaker group 150.

[0038] Referring to Figure 2, the present invention provides a method for optimizing the sound effects inside a passenger vehicle cabin. This method is executed by the aforementioned sound effect optimization system 100 and specifically includes the following processing steps:

[0039] The system first executes a monitoring trigger determination process. This process includes receiving user commands from the human-machine interface module 160 or responding to periodic interrupt signals from the internal timer of the cockpit domain controller 110. When monitoring is triggered, the cockpit domain controller 110 reads vehicle status data through the vehicle communication network and performs an environmental access determination. The environmental access determination logic requires the following physical conditions to be met simultaneously: the vehicle speed is zero, the transmission gear is in parking gear (P), and the powertrain is in a non-drive state. Under the premise that the above vehicle status is met, the system collects the ambient background noise in the cabin through the voice acquisition unit 140 and calculates the sound pressure level of the background noise. Only when the real-time background noise sound pressure level is lower than a preset acoustic threshold, the system locks the window control logic and enters active monitoring mode.

[0040] Upon entering active monitoring mode, the power amplifier 130 retrieves a standard test excitation signal stored in its internal non-volatile storage module 132. This standard test excitation signal is strictly consistent with the signal used to establish the initial reference data, and the signal type is pink noise or a sinusoidal sweep signal covering the entire frequency range from 20Hz to 20kHz. The power amplifier 130 drives the speaker assembly 150 to play the standard test excitation signal at a preset fixed volume level.

[0041] While the excitation signal is playing, the voice acquisition unit 140 samples the sound field inside the cabin in real time, converting the analog sound waves into a digital audio stream and transmitting it to the digital signal processing module 131 of the power amplifier 130 via the digital audio bus 120. The digital signal processing module 131 performs a Fast Fourier Transform (FFT) on the received digital audio stream, converting the time-domain signal into a frequency-domain signal. Subsequently, the digital signal processing module 131 extracts the frequency band according to the 1 / 3 octave or 1 / 2 octave band standard defined by the international standard ISO 266. The real-time sound pressure level values ​​at key center frequency points are used to construct the current state response vector T.

[0042] The digital signal processing module 131 then performs the difference calculation process. This module reads the pre-stored initial reference response vector from the non-volatile memory module 132. and the current state response vector With the initial reference response vector Perform point-by-point difference operations to obtain the spectral deviation vector. .vector Each element in the equation represents the amount of sound pressure level attenuation or drift at the current frequency point relative to the factory setting.

[0043] Based on the calculated spectral deviation vector The system then enters the parameter calculation and adjustment process. The digital signal processing module 131 calls the preset sound field transfer function matrix. The matrix This describes the acoustic contribution of the gain variation of each independent control channel in the speaker array 150 to the sound pressure level at each center frequency point at the location of the voice acquisition unit 140. The digital signal processing module 131 constructs a system of linear equations regarding the gain adjustment of each channel and solves these equations using the least squares method to obtain the target gain compensation value required for each channel in the speaker array 150. .

[0044] After obtaining the target gain compensation value, the power amplifier 130 updates the filter coefficients and channel gain settings of its internal audio processing path and adds the compensation value to the current audio output parameters.

[0045] After parameter adjustment, the system executes a closed-loop verification process. The power amplifier 130 replays the standard test excitation signal and re-acquires sound field data via the voice acquisition unit 140 to generate a verification response vector. The system calculates the verification response vector and compares it with the initial reference response vector. The residual deviation is considered. If the Euclidean norm of the residual deviation is less than the preset convergence threshold, the system determines that the optimization is complete, saves the current gain parameters, and exits the active monitoring mode; if the residual deviation exceeds the convergence threshold, the system repeats the above parameter calculation and adjustment process based on the current residual deviation until the convergence condition is met or the preset maximum number of iterations is reached.

[0046] In the passenger vehicle cabin audio optimization method described in this invention, the monitoring triggering mechanism is integrated into the logic processing unit of the cabin domain controller 110 and configured to generate a request signal to initiate the audio optimization process. This mechanism includes two parallel paths: periodic automatic triggering logic and user-initiated triggering logic, to adapt to different usage scenarios throughout the vehicle's entire lifecycle.

[0047] In the periodic automatic triggering logic, the cockpit domain controller 110 reads the vehicle's cumulative mileage data and the system's cumulative operating time data in real time through the vehicle communication network (such as CAN bus or in-vehicle Ethernet). The non-volatile memory inside the cockpit domain controller 110 records a historical mileage snapshot from the last time the sound optimization was performed. and historical snapshots The logic unit compares the current vehicle mileage with the data at a preset sampling frequency. With historical mileage snapshot The difference, and the current system time With historical snapshots The difference.

[0048] The system has a preset first trigger threshold. (e.g., mileage threshold) and second trigger threshold (For example, a time threshold). When the logic unit determines... or When any one of these conditions is met, the system determines that the vehicle's speaker components have reached the end of their service life where physical performance degradation will occur. At this time, the cockpit domain controller 110 generates a prompt message on the human-machine interface, pushing a sound calibration suggestion to the user. If the user confirms the suggestion through the human-machine interface, the system sets the calibration request flag in the internal status register, thereby generating an automatic trigger request signal.

[0049] In the user-triggered logic, the human-machine interface module 160 of the cockpit domain controller 110 provides a virtual interactive control under the audio management level of the system settings menu. This control is configured to respond to user touch operations. When the human-machine interface module 160 detects a valid touch event for the virtual interactive control, it generates an interrupt signal and sends it to the logic processing unit. The logic processing unit responds to the interrupt signal by directly setting the calibration request flag and generating a user-triggered request signal.

[0050] Whether the request signal is generated periodically and automatically or by the user, it does not directly initiate the test process. Instead, it serves as an input signal for the environmental access judgment logic. Upon detecting the calibration request flag, the cockpit domain controller 110 suspends the current audio entertainment task and invokes the environmental monitoring subroutine to verify whether the vehicle's current physical state meets the test conditions. This logic design, separating the trigger request from the execution action, ensures that monitoring instructions are generated only when the user is aware of or actively requests them, preventing the system from silently initiating high-volume test signals in the background and causing sudden interference to the occupants.

[0051] In the sound effect optimization method described in this invention, the environmental intrusion determination strategy is executed by the cockpit domain controller 110. This aims to establish a test environment with stable acoustic characteristics and a signal-to-noise ratio that meets measurement requirements, thereby ensuring the effectiveness of subsequent frequency response data acquisition. This strategy is configured to execute after receiving a monitoring trigger request signal and before the power amplifier outputs an excitation signal, employing serial or parallel logic to perform multi-dimensional verification of the vehicle's physical state and acoustic environment.

[0052] The cockpit domain controller 110 first performs a vehicle dynamics state verification. The cockpit domain controller 110 reads the vehicle's speed signal and transmission gear position signal via the vehicle controller area network (CAN bus). The logic unit compares the real-time vehicle speed value with zero and simultaneously checks whether the gear position signal is encoded and confirmed to be in parking gear (P). Only when the vehicle speed is zero and the gear is locked in P is the system determined that the vehicle is stationary. This verification logic eliminates broadband interference from tire rolling noise, wind noise, and road surface excitation on the sound field measurement within the cabin.

[0053] After confirming the vehicle is stationary, the system performs a powertrain vibration and noise check. For internal combustion engine vehicles or hybrid vehicles, the cockpit domain controller 110 reads the engine speed signal or start-stop status bit to confirm that the engine is in an OFF state or an automatic start-stop shutdown state; for pure electric vehicles, the system confirms that the electric drive system is in standby mode and the cooling fan is at low speed or off. This logic aims to eliminate the superposition effect of low-frequency mechanical vibration and high-frequency electromagnetic noise generated during powertrain operation on the microphone acquisition signal, especially to prevent measurement distortion in the low-frequency range (such as 20Hz-100Hz) caused by engine idling frequency.

[0054] The system then performs a cabin acoustic sealing test. The cabin domain controller 110 acquires the opening and closing status signals of all doors, windows, sunroof, and trunk lid through the body control module (BCM). The logic unit performs a logical AND operation on the status bits of all the above-mentioned opening and closing components. If any window or door is detected to be not fully closed, the system determines that there is a risk of sound leakage in the current acoustic environment, which may cause external environmental noise intrusion or abnormal attenuation of low-frequency sound pressure levels inside the vehicle due to leakage. The system then terminates the test process and prompts the user to close the corresponding component through the human-machine interface module.

[0055] With the physical state verifications all passed, the system performs active background noise monitoring and verification. The cockpit domain controller 110 instructs the voice acquisition unit 140 to open the acquisition channel and continuously acquire an ambient audio stream of a preset duration (e.g., 3 to 5 seconds) without playing any test excitation signal. The digital signal processing module calculates the root mean square (RMS) amplitude or A-weighted sound pressure level of this ambient audio stream and compares it with a preset background noise threshold (e.g., 45 dB).

[0056] This background noise verification logic is used to identify transient or steady-state acoustic interference that vehicle sensors cannot detect, including but not limited to sudden horn sounds outside the vehicle, noise from raindrops hitting the vehicle body, or conversations between occupants inside the vehicle. If the calculated ambient background noise value is lower than the threshold, the system determines that the current environment meets the signal-to-noise ratio requirements, generates an environmental access confirmation signal, and allows the power amplifier to start playing the standard test excitation signal; if the ambient background noise value is higher than the threshold, the system determines that the current ambient acoustic interference is too great, automatically suspends the test process and enters a waiting retry state or provides feedback to the user that the environment is too noisy. Through the above-mentioned multi-level logic gate control, this invention ensures that the sound effect optimization algorithm only operates in a stable, quiet, and controlled sound field environment.

[0057] In the sound effect optimization method described in this invention, the initial benchmark establishment process is configured to lock the standard acoustic characteristics of the vehicle in its factory state or model development finalization state, thereby providing a comparison anchor point for subsequent aging monitoring.

[0058] The establishment of this initial baseline data is strictly limited to the stage when the physical state of the vehicle's interior environment is finally locked. Specifically, the material distribution and geometry of interior components, including seat materials, carpet installation, headliner, door panels, and dashboard surfaces, constitute an acoustic cavity boundary condition. The reflection, diffraction, and absorption characteristics of this acoustic cavity boundary condition directly determine the frequency response characteristics of the sound waves emitted by the speakers transmitted to the voice acquisition unit 140. Therefore, the establishment of the initial baseline must be carried out after the entire vehicle interior is assembled and in a static, controlled acoustic environment (such as a semi-anechoic chamber or a silent testing room) to eliminate interference from external environmental noise and changes in the interior condition on the consistency of the baseline data.

[0059] During the benchmark establishment process, the digital signal processing module 131 inside the power amplifier 130 generates a standard test excitation signal. The excitation signal The signal is preset to cover the audible range of 20Hz to 20kHz, specifically using either pink noise with uniform energy density or a logarithmically swept sine wave. The power amplifier 130 converts this digital signal into an analog drive voltage, which simultaneously drives all channels or target channels of the speaker group 150 to produce sound.

[0060] The voice acquisition unit 140 simultaneously acquires acoustic response signals within the vehicle cabin. Although the frequency response curve of the in-vehicle voice microphone exhibits non-flat characteristics (e.g., low-frequency roll-off or high-frequency peaks), this invention utilizes the data acquired by this microphone to establish a benchmark, aiming to leverage the hardware consistency of the same physical sensor. As long as the same microphone is used in subsequent monitoring processes, the sensor's own transfer function becomes a fixed component in the system transfer function, mathematically eliminated in subsequent differential calculations, thus eliminating the need to rely on a high-precision measurement-grade microphone.

[0061] The acquired analog signal is converted to digital signal by analog-to-digital converter (ADC) and then enters digital signal processing module 131. Digital signal processing module 131 applies a Hanning window or Hamming window to the time-domain signal to reduce spectral leakage, and then performs a Fast Fourier Transform (FFT) to obtain frequency domain data. The processing module extracts the frequency domain data according to the 1 / 3 octave band filter bank specified in the international standard IEC 61260. The sound pressure level values ​​at key center frequencies. These center frequencies are typically selected as follows: 20Hz, 31.5Hz, 63Hz, 125Hz, 250Hz, 500Hz, 1000Hz, 2000Hz, 4000Hz, 8000Hz, 16000Hz, and 20kHz, which can fully cover the low-frequency, mid-frequency, and high-frequency regions.

[0062] Digital signal processing module 131 will extract the above-mentioned signals. The sound pressure level values ​​at each center frequency point are used to construct the initial reference response vector. This vector The mathematical expression for it is defined as:

[0063] ;

[0064] in, Indicates the first The sound pressure level at each center frequency point under initial standard conditions, in decibels. .

[0065] Once constructed, this initial baseline response vector The data is embedded in a protected area of ​​the non-volatile storage module 132 (e.g., EEPROM or Flash memory) of the power amplifier 130. This storage area is configured to be read-only or can only be overwritten in engineering mode to prevent the loss of reference data due to power failure or system reset during subsequent vehicle use. The ideal frequency response state of the audio system at zero aging moment is defined, which constitutes the numerical origin of all subsequent aging judgment and compensation algorithms.

[0066] In the sound effect optimization method described in this invention, the selection and playback of the excitation signal is the physical basis for ensuring comparability from the initial reference state to the aging monitoring state. This step is precisely controlled by the digital signal processing module 131 inside the power amplifier 130, aiming to provide the speaker group 150 with an acoustic excitation source that has deterministic spectral characteristics and stable energy.

[0067] In this embodiment, pink noise is used as the standard test excitation signal for signal selection. The power spectral density (PSD) of pink noise is inversely proportional to its frequency, i.e. This signal characteristic results in a constant energy distribution on the logarithmic frequency axis. Since the spectrum analysis stage of this invention is based on 1 / 3 or 1 / 2 octave band filter banks as defined in the international standard IEC 61260 (i.e., the bandwidth widens logarithmically with increasing frequency), pink noise contains equal energy within each octave band. This physical characteristic ensures that the ideal frequency response curve is a flat straight line when using the real-time spectrum analysis (RTA) algorithm, thereby significantly reducing the computational load on the digital signal processing module 131 for calculating spectrum deviation and improving the performance for bandwidth attenuation. The calculations offer intuitiveness and accuracy. Alternatively, the system can employ a logarithmic sinusoidal sweep signal that varies continuously from 20Hz to 20kHz over time, suitable for scenarios requiring simultaneous measurement of system nonlinear distortion (THD).

[0068] Regarding the source of the excitation signal, this signal is not input through external devices (such as a mobile phone via Bluetooth or a USB audio source), but is directly generated internally by the power amplifier 130 or the cockpit domain controller 110. Specifically, the digital signal processing module 131 integrates a signal generator algorithm unit or directly stores uncompressed high-sampling-rate (e.g., 48kHz or 96kHz) PCM format test audio files. This internal generation mechanism eliminates encoding / decoding losses, Bluetooth transmission bandwidth limitations, or analog line noise interference introduced by external audio source input, ensuring the digital domain excitation signal source remains consistent throughout the vehicle's entire lifecycle. Maintain bit-perfect consistency.

[0069] In terms of playback control logic, to construct a standardized test path, the system performs strict audio channel takeover and bypass operations. When entering monitoring mode, the digital signal processing module 131 cuts off the audio stream from the entertainment system and forcibly bypasses all user-defined sound effect settings. These user-defined sound effect settings include, but are not limited to, bass, treble, and balance / fader parameters adjusted by the user in the human-computer interaction interface, as well as preset EQ sound effect modes (such as jazz and rock). The system establishes a direct link from the signal generator to the digital-to-analog converter (DAC) to ensure that the drive signal received by the speaker reflects only the physical response of the hardware itself, rather than the result modified by software.

[0070] Furthermore, the output gain control of the excitation signal follows an absolute amplitude lock-in principle. The system does not rely on the user's currently set master volume knob position, but instead forcibly locks the output gain of the digital signal processing module 131 to a preset calibration value (e.g., -20 dBFS). This calibration value is determined during the initial reference establishment phase and ensured to be stored in the non-volatile memory module 132. The subsequent amplifier circuit of the power amplifier 130 (typically a Class D amplifier) ​​drives the speaker assembly 150 according to the pulse width modulation (PWM) signal corresponding to this fixed digital gain. This constant gain control mechanism ensures that the electrical power load on the speaker voice coil is consistent during each test, thereby ensuring that the sound pressure level variation acquired by the voice acquisition unit 140 (…) The change in electroacoustic conversion efficiency is entirely attributable to the change in compliance of loudspeaker mechanical components (such as surrounds and centering plates) or the degradation of magnetic circuit performance, rather than fluctuations in the amplitude of the input signal.

[0071] Referring to Figures 1 and 2, in the sound effect optimization method of the present invention, the sampling and frequency response analysis process is performed by the digital signal processing module 131 inside the power amplifier 130. Its core function is to convert the non-steady-state time-domain acoustic signal picked up by the voice acquisition unit 140 into a quantifiable steady-state frequency-domain feature vector, i.e., the current state response vector. At the front end of the data acquisition link, the voice acquisition unit 140 integrates an analog-to-digital converter (ADC). This ADC is configured to operate at a fixed sampling frequency. (In this embodiment, the frequency is set to 48kHz) The analog sound pressure signal is quantized to generate a digital audio stream in Pulse Code Modulation (PCM) format. This digital audio stream is transmitted to the input buffer of the power amplifier 130 via the digital audio bus 120. To ensure that the effective bandwidth of the frequency analysis covers the audible range of 20Hz to 20kHz, the sampling frequency is... Strictly follow the Nyquist sampling theorem, and set the quantization depth to at least 16 bits or 24 bits to ensure sufficient dynamic range.

[0072] The digital signal processing module 131 reads the digital audio stream from the input buffer and performs framing and windowing preprocessing on it. To capture the fine spectral characteristics of the low-frequency range (especially the 20Hz to 100Hz region), the processing module cuts the continuous audio stream into segments of length [missing information]. Data frames, where the frame length is The resolution is set to 4096 or 8192 points to obtain a sufficiently high frequency resolution. To eliminate the spectral leakage effect caused by time-domain truncation, each frame of data is fitted with a Hanning window function. Perform time-domain multiplication:

[0073] ;

[0074] in, The original input signal, This is the windowed signal. Subsequently, the digital signal processing module 131 processes the windowed signal. Perform a Fast Fourier Transform (FFT) to calculate the discrete spectrum. To obtain stable frequency response characteristics and eliminate the random fluctuations inherent in pink noise, the system employs a linear or exponential averaging strategy to smooth the power spectral density over multiple consecutive time frames (e.g., accumulating 3 to 5 seconds of data) to obtain the average power spectrum. .

[0075] In the frequency band division and energy integration stages, this embodiment strictly follows the 1 / 3 octave band filter specification defined by the International Electrotechnical Commission standard IEC 61260. The digital signal processing module 131 has a pre-set frequency mapping table, which defines... Center frequency points and their corresponding upper and lower cutoff frequencies and For each selected center frequency point The processing module will process the average power spectrum. Integrate (or sum) the energy within the corresponding frequency range to calculate the total energy of that frequency band.

[0076] ;

[0077] in, and Corresponding to cutoff frequencies and Discrete index values ​​in the FFT spectrum. Based on the calculated total energy of the frequency band. Digital signal processing module 131 calculates the... Sound pressure level at each center frequency (in dBFS or relative decibels):

[0078] ;

[0079] The system iterates through and calculates all of the above steps. The sound pressure level at each monitoring frequency point is used to construct the current state response vector. In this embodiment, to balance computational accuracy and system resource consumption, the selected set of center frequency points covers the key sensitive frequency bands of speaker aging, specifically including: 31.5Hz, 40Hz, 50Hz, and 63Hz in the low-frequency band (corresponding to the sensitive area of ​​change in surround compliance), 500Hz and 1kHz in the mid-frequency band (corresponding to the fundamental frequency range of human voice), and 8kHz, 10kHz, and 12.5kHz in the high-frequency band (corresponding to the sensitive area of ​​change in voice coil inductance and diaphragm quality). This feature extraction method based on standard octave bands compresses thousands of FFT frequency point data into feature vectors containing dozens of key values, providing standardized and low-dimensional input data for subsequent transfer function calculation.

[0080] The relative measurement mechanism employed in this invention aims to utilize the hardware consistency of the vehicle-mounted voice acquisition unit 140 to eliminate the interference of the sensor's own non-ideal frequency response characteristics on aging determination. In the acoustic signal transmission link, the digital audio signal acquired by the voice acquisition unit 140 is not a single sound source signal, but rather the superposition result of the transmission functions of the entire electroacoustic link. For any center frequency point... The sound pressure level value output by the voice acquisition unit 140 It consists of a linear superposition of the following components: the level of the internal excitation signal source. Gain of power amplifiers and circuit transmission links Electroacoustic conversion response of loudspeaker group Sound field transfer function of the cabin acoustic environment and the frequency response of the voice acquisition unit itself During the construction of initial baseline data (step S1), the system records the baseline response vector. elements in This corresponds to the speaker being in its initial, unaged state. The overall system response under ().

[0081] During real-time status monitoring, the system records the current status response vector. elements in Corresponding to the speaker being in its current aging state ( The overall system response under ().

[0082] When performing difference calculations, the digital signal processing module 131 ensures the consistency of measurement conditions based on the environmental access determination and playback control logic of this invention. Specifically: excitation signal source For the same digital file, it is guaranteed Unchanged; the power amplifier locks its output gain, ensuring... Unchanged; the environmental access strategy ensures that the windows are closed and the vehicle is stationary, thus maintaining the acoustic environment of the cabin. Keep it within a stable range.

[0083] Most importantly, although the vehicle-mounted voice acquisition unit 140 is a non-measurement-grade microphone, it exhibits sensitivity roll-off or frequency response unevenness in the low or high frequency ranges (i.e., (While not a constant value across the entire frequency band), since the same physical microphone is used for both baseline establishment and real-time monitoring, and the physical characteristics of this microphone degrade very little throughout the vehicle's lifespan, therefore... This is considered a fixed systematic error in this measurement system. When the digital signal processing module 131 calculates the spectral deviation... At that time, the above-mentioned invariant system components ( , , In subtraction, they cancel each other out. Calculation result Only the changes in the physical properties of the speaker components over time were retained, namely:

[0084] ;

[0085] Through this relative measurement logic, the present invention transforms the measurement of absolute sound pressure level into the measurement of sound pressure level change. Even if the absolute sensitivity of the voice acquisition unit 140 is low at certain frequencies (such as 20Hz) and cannot accurately reflect the true sound pressure level at that frequency, as long as the microphone can respond to changes in sound pressure, the system can accurately capture the low-frequency response attenuation trend of the speaker caused by the aging of the suspension system (such as stiffening of the surround and fatigue of the centering support), thereby achieving high-precision aging monitoring based on ordinary vehicle microphones.

[0086] In the sound effect optimization method described in this invention, the construction of the spectral deviation vector is a key intermediate link connecting signal domain analysis and control domain decision-making, and is executed by the digital signal processing module 131 inside the power amplifier 130.

[0087] After completing the sampling and frequency response analysis steps, the current state response vector is temporarily stored in the volatile memory (RAM) of the digital signal processing module 131. It contains The system measures the sound pressure level at the center frequency point in real time. Simultaneously, the digital signal processing module 131 retrieves the initial reference response vector from the non-volatile storage module 132. To ensure dimensional consistency in data operations, the system first processes vector T and vector... Perform an integrity check to confirm the frequency point indexes of both. Strictly correspondence.

[0088] The digital signal processing module 131 performs point-by-point differential operations to quantify the performance drift of the loudspeaker assembly 150 at various characteristic frequency points. A spectral deviation vector is defined. For one A 3D column vector, whose calculation logic is as follows:

[0089] ;

[0090] in, For the first The sound pressure level at each center frequency point under reference conditions. This represents the measured sound pressure level at the same frequency point under the current aging condition.

[0091] Under this mathematical definition, the deviation value The symbols have a clear physical orientation: when When the measured sound pressure level is lower than the reference sound pressure level, it indicates that there is response attenuation at that frequency (e.g., loss of low-frequency sensitivity due to decreased compliance of the speaker suspension system). Positive gain compensation needs to be applied to the system subsequently. If the measured sound pressure level is higher than the reference sound pressure level, it indicates that there is an abnormal gain spike or measurement noise interference at this frequency point. The system needs to apply negative gain attenuation or maintain the status quo.

[0092] To prevent random noise during the measurement process (such as quantization error, thermal noise, or transient micro-airflow) from causing unnecessary minor fluctuations in the control system, this embodiment introduces dead-zone threshold determination logic during the deviation vector construction process. The system has a preset noise tolerance threshold. (For example, 0.5dB). The digital signal processing module 131 generates the final bias vector used for solution. Previously, the original difference results Perform filtering: If If the change at that frequency point is determined to be within the normal measurement error range, then a forced command is executed. ;like Then retain the deviation value and let .

[0093] This threshold determination logic ensures that the system only compensates for statistically significant physical attenuation, avoiding system parameter oscillations caused by hypersensitive regulation.

[0094] Furthermore, considering the differences in human ear sensitivity to loudness changes across different frequency bands, and the different physical aging mechanisms of loudspeakers across different frequency bands, this embodiment also supports the adjustment of the deviation vector. Perform weighted preprocessing. The system defines a... diagonal weighted matrix diagonal elements Representing the The compensation weighting coefficients for each frequency point. For the lowest frequency range of speaker aging (e.g., 20Hz-100Hz) and the high frequency range (e.g., above 10kHz), larger weighting coefficients are set. For the mid-frequency band, set a unit weight ( The bias vector after weighting. This will serve as the direct input to the subsequent least squares solver, thereby guiding the optimization algorithm to prioritize correcting the frequency bands that have the greatest impact on the listening experience or the most severe physical attenuation.

[0095] Referring to Figure 3, the sound effect optimization method of this invention utilizes this mathematical model to describe the mapping relationship between each independent control channel of the speaker assembly 150 and the spectral response measured by the voice acquisition unit 140, thereby providing a deterministic mathematical basis for the decoupled calculation of multi-channel parameters. This sound field transfer function matrix... It is pre-stored in the non-volatile memory module 132 of the power amplifier 130. Matrix Defined as a dimension A real matrix, where This represents the total number of frequency points at the monitoring center (corresponding to the aforementioned spectral deviation vector). (dimensions) This represents the total number of independently gain-adjustable physical channels in the speaker group 150. In this matrix, each column vector corresponds to a speaker channel (e.g., the front left bass channel), and each row vector corresponds to a center frequency point (e.g., ...). .

[0096] matrix any element in It has a clear physical meaning, and it represents the first The gain change of the first speaker channel affects the... The contribution sensitivity of sound pressure level changes at each center frequency point. Specifically, numerical... Quantified when only the first When the output gain of each channel increases by 1dB, the voice acquisition unit 140 in the [missing information]... The increase in sound pressure level (dB) detected at each frequency point. This matrix model is constructed based on the superposition principle of linear time-invariant (LTI) systems. Within a small gain adjustment range (e.g.) Within this system, a linear relationship is assumed between the logarithmic adjustment of the gain of each channel and the logarithmic change of the final sound pressure level. Based on this, the present invention establishes the following core linear equation model:

[0097] ;

[0098] in: The result of the calculation in the previous step The spectral deviation vector is a known quantity (observation value) in the system of equations; for The sound field transfer function matrix is ​​the coefficient matrix (system parameters) of the system of equations; For the solution to be found The dimensional gain adjustment vector, whose elements Representing the The gain compensation value that needs to be applied to each channel. Regarding the matrix... The system's construction originates from a single-channel independent excitation method obtained during the vehicle development or production line calibration phase. Specifically, under standard acoustic conditions, the system sequentially and individually activates each speaker channel. Play broadband test signals and record voice acquisition unit 140 at all The response value at the nth frequency point, thus filling the matrix at the nth frequency point. List.

[0099] It is worth noting that the matrix The sparse distribution of the numerical values ​​objectively reflects the crossover design architecture of the car audio system. For example, in the column vector corresponding to the channel connecting the tweeter, the row element values ​​for the low-frequency band (e.g., 20Hz-200Hz) are close to zero, indicating that adjusting the gain of the tweeter channel will not affect the sound pressure level of the low-frequency band; conversely, in the column vector corresponding to the subwoofer channel, the row element values ​​for the high-frequency band are extremely small. By introducing this transfer matrix that includes the physical crossover characteristics, this invention can achieve automatic decoupling of frequency components and physical channels in the subsequent solution process, ensuring that the spectral deviation in the low-frequency band is compensated only by the subwoofer channel, avoiding erroneous overall gain enhancement across all frequency bands.

[0100] In the sound effect optimization method described in this invention, the inverse problem solving step is executed by the digital signal processing module 131 inside the power amplifier 130. Its core task is to convert the spectral deviation vector in the acoustic domain into a digital signal processing module 131. Channel gain adjustment vector converted to the electrical domain Due to the number of monitoring frequency points Typically much larger than the number of independent control channels for speakers. (Right now This mathematical problem is essentially a problem of solving an overdetermined system of equations, and there is usually no exact solution that can simultaneously reduce the deviations at all frequency points to zero. Therefore, this embodiment uses the least squares method combined with regularization constraints to seek the optimal solution that minimizes the system energy error.

[0101] Digital signal processing module 131 constructs the objective function This function aims to minimize the sum of squared Euclidean distances between the actual compensation effect and the target deviation. Considering the numerical stability of the matrix inversion process and to prevent gain oscillations caused by overfitting, this embodiment introduces a Tikhonov regularization term into the objective function. The objective function is defined as follows:

[0102] ;

[0103] in, For the aforementioned The sound field transfer function matrix is ​​D; D is 3D spectral deviation vector; For the solution to be found The gain adjustment vector is denoted by λ, which is a non-negative regularization parameter used to balance the minimization of residuals with the norm of the solution, preventing the calculated gain adjustment from being too large.

[0104] Based on the above objective function, the digital signal processing module 131 solves the problem through matrix operations. The analytical solution is given by the following formula:

[0105] ;

[0106] In this formula, Representation matrix The transpose of the matrix; The dimension is The identity matrix; This represents the matrix inversion operation; the calculation result. It is a collection of A vector of elements ,in Corresponding to the The theoretically recommended gain adjustment value for each speaker channel, in decibels (dB).

[0107] To ensure that the optimization results do not exceed the physical limits of the speaker assembly 150 or cause the power amplifier 130 to enter clipping mode, the digital signal processing module 131 obtains the theoretically recommended values. Afterwards, safety boundary constraint processing must be performed. The system has a preset set of asymmetric gain adjustment thresholds, namely the positive maximum compensation threshold. (e.g., +3dB) and negative maximum attenuation threshold (e.g., -6dB).

[0108] Digital signal processing module 131 pairs of vectors Each element in Execute the following saturation logic judgment: If Then a mandatory order like Then a mandatory order like Then keep The numerical value remains unchanged. This constraint logic ensures that the system will not output excessive driving energy due to extreme measurement errors at certain frequencies (such as deep valleys caused by standing wave cancellation), thereby protecting the speaker voice coil from burning out.

[0109] After the above calculations and constraint processing, the digital signal processing module 131 will finally determine... The value is applied to the control path of the audio signal stream. Specifically, the digital signal processing module 131 reads the current base gain register values ​​for each channel. and perform update operations. Updated gain parameters It is then written into the audio processing coefficient register of the DSP, which changes the amplitude of the audio signal output to the speaker group 150 in real time, thereby completing the compensation for the aging acoustic characteristics at the physical level.

[0110] After completing the calculation of the theoretical gain parameters, the sound effect optimization method described in this invention does not end the process directly, but enters a rigorous parameter execution and closed-loop verification stage to eliminate control deviations caused by nonlinear factors and ensure the convergence of the final listening effect.

[0111] During the parameter adjustment execution phase, the digital signal processing module 131 inside the power amplifier 130 receives the channel gain adjustment vector after boundary constraint processing. To avoid audible popping or zipper noise at the audio output due to abrupt changes in gain parameters, the digital signal processing module 131 employs a smooth transition strategy to update hardware registers. Specifically, the system sets a transition time window (e.g., 200 milliseconds) and linearly or logarithmically interpolates the current base gain value towards the target gain value, updating the filter coefficients or gain controller register values ​​in small steps. This soft-switching mechanism ensures that the entire calibration process is acoustically smooth for the user, even in a quiet vehicle cabin environment.

[0112] After the gain parameters are updated, the system immediately initiates the closed-loop verification process. Power amplifier 130 reactivates the standard test excitation signal source, driving speaker assembly 150 to play pink noise and instructing voice acquisition unit 140 to perform a new round of sound field data acquisition. Digital signal processing module 131 extracts the verified current state response vector according to the aforementioned frequency response analysis method. .

[0113] Subsequently, the system performs residual evaluation calculations. The digital signal processing module 131 reads the initial reference response vector. Calculate its response vector with the current state after verification. The residual deviation vector between To quantify the overall optimization effect, the system calculates the Euclidean norm of the residual deviation vector. Norm or weighted root mean square error (RMSE) generates a scalar convergence metric.

[0114] The system will use the convergence index With the preset convergence threshold The comparison is performed. This comparison logic forms a branch of a negative feedback control loop.

[0115] Decision point: If This indicates that the current sound field frequency response characteristics have been highly restored to the factory standard, and the system determines that the optimization is successful. At this time, the digital signal processing module 131 triggers the parameter solidification command, writes the updated final gain parameters to the partition of the non-volatile storage module 132 (such as Flash or EEPROM), overwrites the old operating parameters, and sends a calibration completion status code to the cockpit domain controller 110, then exits the monitoring mode and releases the audio channel control.

[0116] like This indicates that even after one compensation, the sound field characteristics still did not reach the expected standard range. This situation stems from the speaker's nonlinear distortion and the sound field transfer matrix. The system checks the current iteration counter value. If the number of iterations has not reached the preset upper limit (e.g., 3 times), the system will use the current residual deviation vector. As new input, the inverse problem-solving algorithm is called again to calculate the correction value of the correction amount, and the next round of fine-tuning iteration is carried out.

[0117] If the number of iterations has reached the upper limit and If the signal still fails to converge, the system will trigger a safety circuit breaker. This typically indicates irreversible physical damage to the speaker components (such as a burned-out voice coil or a broken cone) or a microphone malfunction, which simple gain compensation cannot fix. In this case, the system abandons the current aggressive compensation parameters, automatically rolls back to the last successful historical parameters or factory default parameters, and generates a corresponding diagnostic trouble code (DTC) through the on-board diagnostic (OBD) system, prompting the user that the audio hardware needs repair, thereby preventing the system from falling into an invalid infinite loop or causing further hardware damage due to overdrive.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the sound effects inside a passenger vehicle cabin, characterized in that, Includes the following steps: The initial reference response vector stored in the non-volatile storage module is obtained. The initial reference response vector characterizes the ideal frequency domain acoustic features received at the voice acquisition unit by the speaker group driven by the standard test excitation signal when the vehicle is in the factory standard interior state. When the monitoring trigger conditions are met and the cabin acoustic environment meets the access standards, the standard test excitation signal generated inside the control power amplifier drives the speaker group to emit sound, and the voice acquisition unit collects real-time acoustic signals in the cabin. After frequency domain analysis and processing, a current state response vector is constructed. Based on the difference between the initial reference response vector and the current state response vector, a spectrum deviation vector is constructed. The elements in the spectrum deviation vector represent the relative sound pressure level drift of each center frequency point in the time dimension. A preset sound field transfer function matrix model is called, and the channel gain adjustment vector is calculated through an inverse problem solving algorithm using the spectrum deviation vector as the input observation value. The channel gain adjustment vector contains the target gain compensation value of each independent control channel in the speaker group. The audio processing parameters of the power amplifier are updated according to the channel gain adjustment vector to compensate for the physical aging characteristics of the speaker group.

2. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 1, characterized in that, The process of establishing the initial reference response vector includes: in a controlled acoustic environment where the physical state of the vehicle interior components is locked, using the digital signal processing module of the power amplifier to generate pink noise with uniform energy density as the standard test excitation signal; locking the output gain of the standard test excitation signal to a preset calibration value, driving the speaker group to emit sound through all channels; acquiring the time-domain signal through the voice acquisition unit, and performing windowing, fast Fourier transform, and smooth averaging processing on the time-domain signal; extracting the sound pressure level values ​​of multiple key center frequency points according to the preset octave band filter bank standard, forming the initial reference response vector, and storing it.

3. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 1, characterized in that, The determination and signal control logic for the cabin acoustic environment meeting the access standards includes: acquiring real-time vehicle status data, confirming that the windows are completely closed, the vehicle is stationary, and the background noise level inside the vehicle is below a preset threshold; cutting off the external audio stream input of the in-vehicle entertainment system and bypassing the user-defined sound effect settings, and establishing a direct link from the internal signal generator to the digital-to-analog converter; wherein, during the acquisition of the current state response vector, the type, generation source, and output gain calibration value of the standard test excitation signal are strictly consistent with those when establishing the initial reference response vector.

4. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 1, characterized in that, The steps for constructing the current state response vector include: performing analog-to-digital conversion on the analog signal output by the voice acquisition unit at a sampling frequency that conforms to the Nyquist sampling theorem; dividing the converted digital signal into data frames of fixed length and applying a Hanning window or Hamming window function for preprocessing; calculating the power spectral density of the windowed signal and mapping it to the 1 / 3 octave band division system defined by the International Electrotechnical Commission (IEC) standard; integrating the energy within each band to obtain the sound pressure level value at the corresponding center frequency point, thereby generating the current state response vector with the same dimension as the initial reference response vector.

5. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 1, characterized in that, The step of constructing the spectral deviation vector further includes dead zone threshold determination and weighted preprocessing: calculating the original difference between the initial reference response vector and the current state response vector at the corresponding frequency points; If the absolute value of the original difference value is less than the preset noise tolerance threshold, the deviation value of the frequency point is set to zero; otherwise, the original difference value is retained. The processed difference value is weighted using a preset diagonal weighting matrix, which sets weight coefficients greater than unit values ​​for the low-frequency and high-frequency bands, thereby obtaining the final spectrum deviation vector used for solving.

6. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 1, characterized in that, The sound field transfer function matrix model is a pre-calibrated real matrix. The number of rows in the real-time matrix corresponds to the total number of monitoring center frequency points, and the number of columns corresponds to the total number of independent control channels in the loudspeaker group. The numerical values ​​of the elements in the matrix represent the sensitivity of the sound pressure level change at the center frequency caused by the unity gain change of the loudspeaker channel; the numerical distribution of the matrix includes the crossover characteristics of the loudspeaker system, which is used to achieve automatic decoupling of frequency components from physical control channels during the calculation process.

7. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 6, characterized in that, The steps for calculating the channel gain adjustment vector using the inverse problem solving algorithm include: constructing an objective function aimed at minimizing system energy error, the objective function including the squared Euclidean distance term between the actual compensation effect and the target deviation, and a Tikhonov regularization term used to limit the norm of the solution; analytically solving the objective function using the least squares method to calculate the theoretically proposed gain adjustment value; performing safety boundary constraint processing on the theoretically proposed gain adjustment value, if the value exceeds a preset positive maximum compensation threshold or negative maximum attenuation threshold, then forcibly truncating the theoretically proposed gain adjustment value to the corresponding threshold boundary to obtain the final channel gain adjustment vector.

8. The method for optimizing the sound effects inside a passenger vehicle cabin according to claim 1, characterized in that, It also includes parameter adjustment execution and closed-loop verification steps: using a smooth transition strategy, the channel gain adjustment vector is superimposed on the current base gain register of the power amplifier; the speaker group is driven again to play the standard test excitation signal and the feedback signal is collected, and the verified residual deviation vector is calculated; it is determined whether the convergence index of the residual deviation vector is less than the preset convergence threshold: if yes, the updated parameters are solidified to the non-volatile storage module; if no and the maximum number of iterations has not been reached, the inverse problem is re-executed with the residual deviation vector as input; if no and the maximum number of iterations has been reached, the parameter state before adjustment is rolled back and a fault diagnosis code is generated.

9. A method for optimizing the sound effects inside a passenger vehicle cabin according to claim 2, characterized in that, The standard test excitation signal is directly generated by the digital signal processing module inside the power amplifier through an algorithm or obtained by reading a pre-stored high sampling rate pulse code modulation file. The standard test excitation signal maintains bit-level consistency in the digital domain and is not input through an external transmission line.

10. A device for optimizing the sound effects inside a passenger vehicle cabin, characterized in that, A method for optimizing in-cabin audio effects in a passenger vehicle as described in any one of claims 1-9, comprising: a storage module for storing an initial reference response vector, a sound field transfer function matrix model, and a preset control threshold; an environmental monitoring and access module for acquiring vehicle status information, determining whether the cabin acoustic environment meets access standards, and controlling the switching and bypassing of audio paths; a signal processing and analysis module for generating standard test excitation signals to drive speaker groups, processing acoustic signals acquired by a voice acquisition unit, and constructing a current state response vector and a spectral deviation vector; a calculation and optimization module for solving the channel gain adjustment vector using a least squares algorithm with regularization constraints based on the spectral deviation vector and the sound field transfer function matrix model; and a closed-loop control module for performing smooth updates of gain parameters, residual verification, and final fixation or rollback operations of parameters.

Images