Adaptive equal-loudness curve dynamic loudness compensation method, system, device and storage medium

Abstract

The present application relates to the technical field of digital audio signal processing, and particularly relates to a self-adaptive equal-loudness curve dynamic loudness compensation method, system, device and storage medium, comprising: obtaining a high-density equal-loudness curve matrix established in advance, determining a reference equal-loudness curve, and collecting a current playing volume of an audio device in real time to calculate an estimated sound pressure level; obtaining a current equal-loudness curve from the matrix according to the estimated sound pressure level, calculating a difference curve of the current equal-loudness curve and the reference equal-loudness curve, and calculating a dynamic compensation proportion coefficient based on the estimated sound pressure level and a preset threshold interval, and combining the difference curve to generate a target compensation gain curve; converting the target compensation gain curve into a digital filter coefficient, performing loudness compensation at a preset audio signal link position, and outputting after performing clipping protection verification on the compensated signal. The present application realizes dynamic loudness self-adaptive following in a full volume range, full and balanced frequency response perception at a low volume, and zero clipping distortion throughout the whole process.

Description

Technical Field

[0001] This invention relates to the field of digital audio signal processing technology, and in particular to an adaptive equal-loudness curve dynamic loudness compensation method, system, device and storage medium. Background Technology

[0002] The human ear's perception of sound loudness is not linear, but rather a complex nonlinear relationship that varies simultaneously with frequency and sound pressure level (SPL). The ISO 226:2003 standard, published by the International Organization for Standardization in 2003, systematically describes the equal loudness perception characteristics of the human ear at different SPL levels for various frequencies through extensive subjective listening tests, forming a family of equal loudness curves with loudness level (square) as the parameter. This standard shows that at low SPL levels, the human ear's sensitivity to low frequencies below 1kHz and high frequencies above 8kHz is significantly lower than that to mid-frequency frequencies. For example, at 40dB SPL, the human ear's sensitivity to low frequencies around 100Hz decreases by approximately 15 to 20dB compared to the 83dB SPL reference loudness level, and its sensitivity to high frequencies at 10kHz decreases by approximately 10dB. This means that when a user turns down the playback volume, not all frequency bands become uniformly quieter; rather, low and high frequencies disappear from the perceived sound before mid-frequency frequencies, resulting in a severe imbalance in the perceived sound spectrum.

[0003] Current audio playback devices generally employ a simple full-frequency linear attenuation method for volume control, applying the same proportion of gain attenuation to all frequency bands, completely ignoring the non-linear characteristics of human auditory perception of loudness. In low-volume playback scenarios, users commonly experience a deterioration in sound quality, such as thinness, lack of bass, and dull high frequencies.

[0004] While some high-end audio devices offer loudness function buttons, their compensation curves are typically fixed parameters and do not dynamically adjust with real-time playback volume. In practical use, fixed loudness compensation often results in insufficient compensation at low volumes and overcompensation at medium volumes, leading to a significant disconnect from the user's actual listening experience. Furthermore, when fixed compensation curves are overlaid with other audio processing modules such as equalizers and room acoustic calibration, the lack of a linkage mechanism can easily cause clipping distortion.

[0005] Therefore, there is an urgent need for an adaptive loudness compensation scheme that can sense playback volume in real time, dynamically calculate compensation curves, work in conjunction with other audio processing modules, and has clipping protection mechanisms. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an adaptive equal-loudness curve dynamic loudness compensation method, comprising the following steps: Obtain a pre-established high-density equal-loudness curve matrix and determine a reference equal-loudness curve; Real-time acquisition of the current playback volume of the audio device, and calculation of the estimated sound pressure level corresponding to the current playback scene; The current equal loudness curve is obtained by interpolation from the high-density equal loudness curve matrix based on the estimated sound pressure level, and the difference curve between the current equal loudness curve and the reference equal loudness curve is calculated. The dynamic compensation ratio coefficient is calculated based on the estimated sound pressure level and the preset threshold range, and the target compensation gain curve is generated by combining the difference curve. The target compensation gain curve is converted into digital filter coefficients, and loudness compensation is performed on the audio signal at a preset audio signal link position. Perform clipping protection verification on the compensated audio signal and output the audio signal after successful verification.

[0007] Preferably, the reference equal loudness curve is an equal loudness curve corresponding to the 83rd loudness level; The high-density equal-loudness curve matrix is ​​obtained by interpolating and extending the key frequency point data in the ISO 226:2003 standard using cubic spline interpolation, resulting in high-density data covering the range of 20Hz to 20kHz with a resolution of 1 / 12 octave band.

[0008] Preferably, the dynamic compensation ratio coefficient is calculated based on a preset threshold range, including: The current digital volume control value is read from the audio driver layer at a preset sampling period, and the estimated sound pressure level is calculated in combination with the full-power output sound pressure level of the device; Set an upper threshold and a lower threshold, and calculate the initial compensation ratio coefficient within the linear interpolation interval formed by the upper threshold and the lower threshold; When the estimated sound pressure level is greater than or equal to the upper limit threshold, the initial compensation ratio coefficient is forced to be zero; when the estimated sound pressure level is less than or equal to the lower limit threshold, the initial compensation ratio coefficient is forced to be the full value.

[0009] Preferably, after calculating the dynamic compensation scaling factor, a smoothing filtering mechanism is also included: The initial compensation scaling factor is subjected to a direction-aware dual-time-constant adaptive smoothing filter to obtain the dynamic compensation scaling factor. The smoothing filter process dynamically selects an effective time constant based on the direction of change of the current initial compensation ratio coefficient relative to the previous smoothing coefficient. This allows the dynamic compensation ratio coefficient to adopt a differentiated transition rate in both volume decrease and volume increase scenarios. In this way, while suppressing the perceptible timbre jump caused by abrupt changes in the compensation curve, it improves the timeliness of low-volume compensation establishment and the auditory naturalness of compensation exit when the volume rises.

[0010] Preferably, calculating the difference curve between the current equal-loudness curve and the reference equal-loudness curve, and generating a target compensation gain curve by combining the difference curve, includes: Calculate the sound pressure level difference between the reference equal loudness curve and the current equal loudness curve at each frequency point to obtain the theoretical compensation requirement curve; The theoretical compensation demand curve is partitioned by frequency band, retaining the complete compensation amount of the low frequency band, linearly reducing the compensation amount of the mid frequency band to zero, and restoring the compensation amount of the preset proportion in the high frequency band to obtain the difference curve after frequency band weighting. The target compensation gain curve is generated by multiplying the frequency band weighted difference curve with the dynamic compensation ratio coefficient.

[0011] Preferably, converting the target compensation gain curve into digital filter coefficients includes: A pre-calculated lookup table for the dynamic compensation ratio coefficient and the IIR filter coefficient group is pre-established. The IIR filter coefficient group contains several cascaded Biquad sections, which are used for low-frequency low-profile filtering, mid-frequency peak filtering and high-frequency high-profile filtering, respectively. The pre-calculated lookup table is queried based on the current dynamic compensation ratio coefficient, and the target IIR filter coefficient group is obtained through linear interpolation.

[0012] Preferably, after obtaining the target IIR filter coefficient set, a coefficient smoothing update mechanism is also included: The coefficients of the currently running filter are gradually updated using linear interpolation in the coefficient domain; Within each preset update cycle, the currently running filter coefficients are moved closer to the target IIR filter coefficient group by a preset proportional step size until they converge to the target value, in order to eliminate the zipper noise caused by coefficient jumps.

[0013] Preferably, applying loudness compensation to the audio signal at a preset audio signal link location includes: An adaptive loudness compensation filter, constructed based on the digital filter coefficients, is inserted into the audio processing pipeline; The audio processing pipeline is executed in the following order: input PCM signal, indoor acoustic calibration filtering, user equalizer processing, adaptive loudness compensation filtering, digital volume attenuation control, and output PCM signal.

[0014] Preferably, it also includes a linkage mechanism with preset scenes: In response to the user's scene preset switching command, at least one of the upper limit threshold, lower limit threshold and maximum gain upper limit parameters in the preset threshold range is updated synchronously; The smoothing filter is applied to the updated parameters to avoid sudden changes in timbre during mode switching.

[0015] Preferably, the clipping protection verification includes a look-ahead peak prediction mechanism: Obtain the maximum sample value of the current audio frame and determine the maximum full-band gain corresponding to the current target compensation gain curve; Multiply the maximum sampled value by the linear gain factor corresponding to the maximum gain across the full bandwidth to estimate the maximum signal amplitude that may be achieved after compensation; Determine whether the maximum signal amplitude is greater than a preset safety threshold. If it is, trigger the clipping protection response process.

[0016] Preferably, the clipping protection response process includes a three-level clipping protection strategy, including: Level 1 protection: Automatically reduces global pregain to provide margin for loudness compensation; Level 2 protection: If the maximum signal amplitude still exceeds the limit after implementing Level 1 protection, the dynamic compensation ratio coefficient is reduced proportionally until the maximum signal amplitude is not greater than the safety threshold. Level 3 protection: If the dynamic compensation ratio coefficient is reduced to a preset lower limit and the clipping risk still exists, loudness compensation is temporarily and completely disabled, and then restored after the preset safety release conditions are met.

[0017] Preferably, it also includes an adaptive learning step for clipping protection events: Record the event information for each trigger of the clipping protection response process and store it in the event log. The event information includes the trigger timestamp, audio content type, and dynamic compensation ratio coefficient for final convergence. Analyze the event distribution patterns in the event log within a preset period. If the frequency of a specific audio input source triggering protection at a preset level or higher exceeds the alarm threshold, the upper limit threshold corresponding to that specific audio input source will be automatically lowered to achieve personalized optimization.

[0018] Based on the same concept, the present invention also provides an adaptive equal loudness curve dynamic loudness compensation system, including an equal loudness curve database preprocessing module, a real-time volume perception and sound pressure estimation module, an equal loudness difference compensation curve calculation module, a dynamic compensation gain generation module, a filter coefficient conversion and signal link superposition module, and a clipping protection and safe output module. The equal loudness curve database preprocessing module is used to obtain a pre-established high-density equal loudness curve matrix and determine a reference equal loudness curve; The real-time volume sensing and sound pressure estimation module is used to collect the current playback volume of the audio device in real time and calculate the estimated sound pressure level corresponding to the current playback scene. The equal loudness difference compensation curve calculation module is used to interpolate the current equal loudness curve from the high-density equal loudness curve matrix based on the estimated sound pressure level, and to calculate the difference curve between the current equal loudness curve and the reference equal loudness curve. The dynamic compensation gain generation module is used to calculate the dynamic compensation ratio coefficient based on the estimated sound pressure level and the preset threshold range, and generate the target compensation gain curve in combination with the difference curve. The filter coefficient conversion and signal link superposition module is used to convert the target compensation gain curve into digital filter coefficients and perform loudness compensation on the audio signal at a preset audio signal link position. The clipping protection and safety output module is used to perform clipping protection verification on the compensated audio signal and output the audio signal after successful verification.

[0019] Based on the same concept, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, it implements the steps of the adaptive equal-loudness curve dynamic loudness compensation method as described in the embodiments.

[0020] Based on the same concept, the present invention also provides a computer-readable storage medium, wherein when the computer-readable instructions are executed by one or more processors, the one or more processors perform the steps of the adaptive equal-loudness curve dynamic loudness compensation method as described in any one embodiment.

[0021] Compared with the prior art, the beneficial effects of the present invention are: (1) This invention pre-establishes a high-density equal loudness curve matrix and estimates the sound pressure level by collecting the playback volume in real time. Based on the estimated sound pressure level and the preset threshold range, it calculates the dynamic compensation ratio coefficient and generates the target compensation gain curve by combining the difference curve between the current equal loudness curve and the reference equal loudness curve. This achieves adaptive following where the compensation amount automatically increases as the playback volume decreases and automatically decreases to zero as the volume increases. This completely replaces the traditional fixed curve loudness switch and compresses the user's perceived frequency response deviation from an average of ±12dB to within ±2.5dB in the full volume range of 30dB to 75dB SPL, significantly improving the low-frequency fullness and high-frequency airiness in low-volume scenarios.

[0022] (2) This invention converts the target compensation gain curve into digital filter coefficients and performs loudness compensation on the audio signal at a preset audio signal link position. The audio processing pipeline sequentially performs indoor acoustic calibration, user equalizer processing, adaptive loudness compensation filtering and digital volume attenuation, realizing complete collaboration between the loudness compensation module and the existing audio processing module. It can be integrated without modifying the code of other modules. At the same time, it is combined with a first-order low-pass smoothing filter processing mechanism to make the compensation coefficients transition smoothly during the volume adjustment process, avoiding perceptible timbre jumps and realizing imperceptible adaptive loudness following throughout the process.

[0023] (3) This invention performs clipping protection verification on the compensated audio signal, uses a forward-looking peak prediction mechanism to estimate the maximum signal amplitude that may be reached after compensation, and triggers a three-level clipping protection strategy when a clipping risk is detected, including prioritizing the reduction of global pre-gain, proportionally reducing the compensation ratio coefficient, and temporarily disabling compensation, to ensure zero clipping events in the full volume range. At the same time, it records clipping protection event information and analyzes the distribution pattern. When protection is frequently triggered by a specific audio input source, the corresponding upper limit threshold is automatically lowered to achieve adaptive learning-based personalized optimization, taking into account both listening experience improvement and system security, and has high reliability and flexibility. Attached Figure Description

[0024] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0025] Figure 1 This is a flowchart of an adaptive equal-loudness curve dynamic loudness compensation method according to the present invention; Figure 2 This is a schematic diagram of an adaptive equal-loudness curve dynamic loudness compensation system according to the present invention; Figure 3 This is a schematic diagram of one embodiment of the computer device of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. Obviously, the described embodiments are only some, not all, of the embodiments described in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without creative effort are within the scope of protection of this application.

[0027] Those skilled in the art will understand that, unless otherwise stated, the singular forms “a” and “an” used herein, and “the”, may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0028] First Embodiment Please see Figure 1 As shown, this embodiment provides an adaptive equal-loudness curve dynamic loudness compensation method, applied to audio playback devices, including but not limited to smart speakers, streaming media audio receivers, digital audio players, home theater processors, automotive audio systems, and mobile terminal devices, comprising the following steps: S1: Obtain the pre-established high-density equal-loudness curve matrix and determine the reference equal-loudness curve.

[0029] This step is completed during the initialization phase and aims to load the basic equal loudness curve data, construct a high-density matrix through interpolation expansion, and establish a reference equal loudness curve as the acoustic benchmark for subsequent compensation calculations.

[0030] S101: The reference equal-loudness curves are equal-loudness curves corresponding to 83 loudness levels. Specifically, in this embodiment, during the firmware initialization phase, the key frequency point data of the equal-loudness curves in the ISO 226:2003 standard are preloaded into the device's non-volatile memory (Flash) to form an equal-loudness curve query database. The standard data covers 29 frequency points (20Hz to 12500Hz), with a loudness level range of 20Phon to 90Phon, a step size of 10Phon, and a total of 8 standard equal-loudness curves.

[0031] For example, writing standard data into a Flash lookup table. Using 1kHz, 83Phon as a reference point: , The difference between the two is 20dB, indicating that the lack of low-frequency perception at low volumes is the theoretical basis for loudness compensation.

[0032] S102: The high-density equal-loudness curve matrix is ​​obtained by interpolating and extending the key frequency point data in the ISO 226:2003 standard using cubic spline interpolation to obtain high-density data covering the range of 20Hz to 20kHz with a resolution of 1 / 12 octave band. Specifically, in this embodiment, cubic spline interpolation is used to extend 29 frequency points to cover the range of 20Hz to 20kHz with a resolution of 1 / 12 octave band (a total of 284 frequency points), forming a high-density equal-loudness curve matrix. It is stored in the device's RAM for real-time querying.

[0033] For example, in the 80Hz to 125Hz range, this is expanded to include nine interpolation points: 80, 84, 89, 94, 100, 106, 112, 119, and 126 Hz. For a 40 Phon equal-loudness curve, the interpolation values ​​are... , The interpolation results are smoothly connected to the standard endpoints, eliminating the step distortion of the compensation curve caused by the sparsity of data points.

[0034] S103: Use 83 Phon (corresponding to the professional studio monitoring reference loudness) as the reference loudness level. ,from Extract The corresponding 284 frequency point data points constitute the reference equal-loudness curve. This serves as the sole benchmark for subsequent compensation gain calculations.

[0035] S2: Real-time acquisition of the current playback volume of the audio device and calculation of the estimated sound pressure level corresponding to the current playback scene.

[0036] This step is the sensory input stage for dynamic loudness compensation, which aims to convert the user-adjusted digital volume value into a sound pressure level estimate with physical meaning, providing a continuously changing dynamic input variable for automatic decision-making on compensation intensity.

[0037] S201: Reads the current digital volume control value from the ALSA audio driver layer in real time with a sampling period of 10ms. The value ranges from 0 to 100, where 0 represents silent operation and 100 represents the device at full power.

[0038] S202: Read the factory-calibrated full-power output sound pressure level of the device from the device calibration partition of non-volatile memory. This parameter is defined as the physical sound pressure level measured in dB SPL when the measuring microphone is located 1 meter in front of the device in an anechoic chamber environment, a 1kHz sine wave standard test signal is input, and the digital volume is turned up to the maximum value (100).

[0039] S203: Calculates the estimated sound pressure level for the current playback scene based on the logarithmic mapping relationship between digital audio signal amplitude and physical sound pressure level. : ,like Then directly determine (Mute), in actual implementation, it can be set to -100 dB SPL as the boundary value.

[0040] For example, the acoustic specifications of a certain device. When the user adjusts the volume to (i.e., 30%), calculated as follows The current sound pressure level is determined to be approximately 72.5 dB SPL, corresponding to a loudness level of approximately 72.5 phon. At this point, a perceptible frequency response difference exists between the human ear's perception and the 83 phon reference level, and adaptive loudness compensation should be initiated.

[0041] S3: Based on the estimated sound pressure level, interpolate the current equal loudness curve from the high-density equal loudness curve matrix and calculate the difference curve between the current equal loudness curve and the reference equal loudness curve.

[0042] This step aims to obtain an accurate description of the human ear's equal loudness perception characteristics at the current volume through two-dimensional interpolation of the equal loudness curve matrix, and compare it with the reference curve to quantify the degree of loss in hearing sensitivity at each frequency band.

[0043] S301: Based on the definition that the sound pressure level at 1 kHz is equal to the loudness level, Directly mapped to the current loudness level .

[0044] S302: Find the surrounding area in the standard loudness level sequence (20, 30, ..., 90 Phon). adjacent gears and .like If the value exceeds the matrix range, the boundary level curve is used directly. (This condition must be met.) .like If the value exceeds the matrix range, the curve corresponding to the boundary level is used directly, and the interpolation step is omitted.

[0045] S303: Within the normal range, calculate Normalized position relative to the lower-end trim level The weight value w ranges from [0, 1].

[0046] S304: For each of the 284 frequency points Extracting low-end loudness level curve values ​​from a high-density matrix and high-end loudness level curve value The current equal-loudness curve is synthesized through linear weighting. : .

[0047] For example, when At that time, it is between 70 Phon and 80 Phon, w=0.25. For a frequency of 100Hz... , Insertion is worthwhile This value indicates that at a loudness level of 72.5 phon, the human ear requires a physical sound pressure level of 53.8 dB SPL to perceive a 100 Hz sound.

[0048] S305: For each frequency point Calculate the sound pressure level difference between the reference equal-loudness curve and the current equal-loudness curve to form the theoretical compensation demand difference curve. : . This indicates that the frequency band is perceived as weak at the current volume and needs to be boosted. This indicates a strong perception, which requires attenuation (usually only occurs in extreme cases).

[0049] For example, calculate the compensation requirement for 100 Hz: .when hour, The compensation requirement is relatively small. If the volume is reduced to 40 phons, ,but The 12.2 dB difference is the theoretical gain value that needs to be compensated at 100 Hz under low volume conditions, and it is the acoustic basis for the adaptive compensation method of this invention.

[0050] S4: Calculate the dynamic compensation ratio coefficient based on the estimated sound pressure level and the preset threshold range, and generate the target compensation gain curve by combining the difference curve.

[0051] This step involves dynamic decision-making and curve synthesis for compensation intensity. It aims to establish a continuous linear negative correlation between compensation intensity and playback volume, and generate the final target compensation curve that conforms to the psychoacoustic perception characteristics of the human ear through frequency band partitioning and weighted processing.

[0052] S401: Calculate the dynamic compensation ratio coefficient based on a preset threshold range, including: The current digital volume control value is read from the audio driver layer at a preset sampling period, and the sound pressure level is estimated by combining it with the full-power output sound pressure level of the device. ; An upper and lower threshold are set, and an initial compensation ratio coefficient is calculated within the linear interpolation interval formed by the upper and lower thresholds. Specifically, in this embodiment, an upper threshold is set. = 75dB SPL, lower limit threshold .

[0053] When the estimated sound pressure level is greater than or equal to the upper threshold, the initial compensation ratio is forced to zero; when the estimated sound pressure level is less than or equal to the lower threshold, the initial compensation ratio is forced to the full value. Specifically, in this embodiment, if... Then the initial scaling factor (Completely exit compensation at high volume); If (Apply full compensation at low volume), then ;like Then calculate using linear interpolation. .

[0054] For example, when hour, When the volume drops to approximately 63 dB SPL, The compensation amount automatically increases as the volume decreases.

[0055] Preferably, a smoothing filtering mechanism is included after calculating the dynamic compensation scaling factor: S402: Apply direction-aware dual-time-constant adaptive smoothing filtering to the initial compensation scaling factor to obtain the dynamic compensation scaling factor; The smoothing filter process dynamically selects an effective time constant based on the direction of change of the current initial compensation ratio coefficient relative to the previous smoothing coefficient. This allows the dynamic compensation ratio coefficient to adopt a differentiated transition rate in both volume decrease and volume increase scenarios. In this way, while suppressing the perceptible timbre jump caused by abrupt changes in the compensation curve, it significantly improves the timeliness of low-volume compensation establishment and the auditory naturalness of compensation exit when the volume rises.

[0056] The Direction-Aware Dual-Tau Adaptive Smoothing (DTAS) algorithm is employed in each sampling period. Perform the following three steps: Direction determination and effective time constant selection: Let the initial compensation ratio coefficient obtained by linear interpolation of the threshold interval in the current step be... The smoothing coefficient that has taken effect in the previous step is The direction of change in the compensation amount can be determined based on the relationship between the two values. like ≥ This indicates that the compensation amount tends to increase (corresponding to the scenario where the user lowers the volume), so a shorter Attack time constant is selected. As the current effective time constant This allows the compensation coefficient to be established quickly, shortening the perception window for bass loss; like < This indicates that the compensation amount tends to decrease (corresponding to scenarios where users increase volume), so a longer Release time constant is selected. As the current effective time constant To allow the compensation coefficient to gradually withdraw and avoid auditory abrupt changes caused by sudden contraction of low frequencies, in this embodiment, a setting is provided. , .

[0057] Calculate the smoothing coefficient for the current step. Based on the selected effective time constant and sampling period Calculate the first-order recursive smoothing coefficient : ,because and For fixed preset values, and This can be pre-calculated as a constant during system initialization and then directly looked up and called at runtime. Substitute... The calculation yields: , .

[0058] First-order recursive smoothing output: using the calculated smoothing coefficient For the current initial compensation ratio coefficient With the smoothing coefficient in the previous step By performing a weighted recursive calculation, the dynamic compensation ratio coefficient that will ultimately take effect in the current step is obtained. : .

[0059] For example: the full-power sound pressure level of a certain device Taking an upper limit threshold of 75dB SPL and a lower limit threshold of 45dB SPL as an example.

[0060] Scenario A (rapidly decreasing volume): The user lowers the volume from 75dB SPL to 50dB SPL. It jumps from 0 to 0.833. Because... ≥ The system selects ,correspond After 10 steps (100 ms), Approximately 0.528; after 16 steps (160 ms), It converges to approximately 0.75, reaching 90% of the target value. If the original fixed time constant scheme (e.g., τ=200ms) is used... ), in the same time period With a value of only about 0.45, the compensation setup is significantly delayed, and users will perceive a thin low-frequency response within approximately 200ms.

[0061] Scenario B (volume subsequently increases): The user reduces the volume from 50dB SPL to 75dB SPL. It decreased from 0.833 to 0. Because... < The system selects ,correspond After 50 steps (500 ms), The compensation value decreases to approximately 0.304; it gradually decreases to near zero within about 1150 ms, resulting in a progressively narrowing low-frequency perception that is almost imperceptible to the human ear. If a fixed τ=200 ms scheme is used, the compensation value decreases to near zero in about 690 ms, which is too fast and the user may perceive a noticeable low-frequency "contraction" jump.

[0062] Compared with the original fixed-time-constant first-order low-pass filter, the DTAS algorithm introduced in this embodiment has the following beneficial technical effects: (1) Eliminate the low volume drop perception delay: The Attack path adopts a short time constant of 100ms. The compensation coefficient reaches more than 80% of the target value within about 100 to 160 ms after the user lowers the volume, compressing the perceptible window of "low frequency brief absence" to within the human ear loudness transient perception threshold (about 200 ms). (2) Eliminate low-frequency jumps when volume rises: The Release path uses a 500ms long time constant and the compensation curve exits slowly to avoid a perceptible "sudden contraction of low frequencies" effect when the volume rises, ensuring the continuity of tone throughout the process; (3) Aligned with dynamic processor industry standards: The Attack / Release dual time constant design is completely consistent with the standard engineering practices in the professional dynamic processing field (compressors, limiters), with sufficient theoretical basis and mature engineering verification, and strong parameter interpretability; (4) Zero additional computational overhead: Compared with the original scheme, only one additional direction comparison and table lookup operation is added. The additional computation amount per frame does not exceed 2 multiplication and addition operations. The increase in CPU usage is negligible and it is fully applicable to embedded platforms.

[0063] S403: The theoretical compensation demand curve is partitioned by frequency band, retaining the complete compensation amount in the low-frequency band, linearly reducing the compensation amount in the mid-frequency band to zero, and restoring a preset proportion of compensation amount in the high-frequency band, resulting in a frequency-weighted difference curve. Specifically, in this embodiment, 100% compensation amount is retained in the low-frequency band (20Hz-200Hz); the compensation amount in the mid-frequency band (200Hz-2kHz) is linearly reduced to 0%; and 50%-70% compensation amount is restored in the high-frequency band (2kHz-20kHz) in the 8kHz-12kHz range. The weighted difference curve is obtained. ; S404: Multiply the frequency-band weighted difference curve with the dynamic compensation scaling factor to generate the target compensation gain curve. : .

[0064] Example: The curves after partitioning are as follows: 100Hz +4.9dB, 500Hz +1.2dB, 1kHz +0.3dB, 2kHz 0dB, 8kHz +2.1dB, 16kHz +1.4dB.

[0065] S5: Convert the target compensation gain curve into digital filter coefficients and perform loudness compensation on the audio signal at the preset audio signal link position.

[0066] This step is the physical implementation and signal link integration of loudness compensation. It aims to fit the discrete target gain curve to IIR filter coefficients that can be efficiently executed by a digital signal processor and precisely embed them in the optimal position in the audio processing pipeline.

[0067] S501: During the device manufacturing or firmware compilation stage, the dynamic compensation scaling factor α is uniformly divided into 100 levels [0, 1] (α = 0.00, 0.01, ..., 1.00). For each... The level is determined by using a least-squares fitting algorithm to obtain the corresponding target curve. The parameters are approximated as six cascaded biquad filters. The six biquad sections are divided as follows: Sections 1-2: low-frequency filter, handling low-frequency boost from 20 Hz to 200 Hz; Sections 3-4: peak filter, handling fine-tuning of the mid-frequency fade-out region; Sections 5-6: high-frequency filter, handling high-frequency airiness restoration. The analog domain parameters (center frequency, gain, quality factor Q) of each biquad section are converted into five digital domain coefficients (b0, b1, b2, a1, a2) using a bilinear transform. All 100 groups, with 30 coefficients per group, are stored in non-volatile memory, forming a pre-calculated lookup table. Furthermore, for the various sampling rates supported by the device (e.g., 44.1 kHz, 48 kHz, 96 kHz, 192 kHz), a separate set of biquad coefficient lookup tables is stored for each sampling rate to ensure that loudness compensation maintains consistent frequency response characteristics across different sampling rates.

[0068] S502: In running mode, obtain the current smoothed dynamic compensation ratio coefficient. Locate adjacent levels among 100 discrete levels. and Calculate interpolation weights For each of the 30 coefficients, the corresponding values ​​are extracted from the lookup table, and linear interpolation is performed to obtain the target IIR filter coefficient set. , .

[0069] Example: At that time, after interpolation by looking up the table, the following coefficients were obtained: Section 1 Low-level filter cutoff frequency Gain Quality factor Section 5. Cutoff Frequency of Overhead Filter Gain Quality factor The total computation of cascading 6 Biquad sections is approximately 120 multiply-accumulate operations per sampling point, with a CPU utilization of about 0.3% on embedded platforms (such as A113L2), which has a negligible impact on performance.

[0070] S503: Preferably, after obtaining the target IIR filter coefficient set, a coefficient smoothing update mechanism is also included: The coefficients of the currently running filter are gradually updated using a coefficient-domain linear interpolation method. Specifically, in this embodiment, when... During updates (every 10ms cycle), AGGM performs gradual updates to the IIR coefficients, employing coefficient-domain interpolation to prevent zipper noise. ; Within each preset update cycle, the currently running filter coefficients are moved closer to the target IIR filter coefficient set by a preset proportional step size until they converge to the target value. This eliminates zipper noise caused by coefficient jumps, where γ = 0.01 (1% of the target difference is updated every 10ms). This ensures a smooth transition of coefficients from the old to the new value, with a transition time of approximately 100ms, well below the human ear's perception threshold for timbre changes (approximately 200ms). The user quickly adjusts the volume from 50% (α=0.3) to 20% (α=0.6). In the following 100ms (10 update cycles), the system smoothly and linearly transitions the Biquad 1's low-frequency gain from +3.6dB to +7.2dB, increasing by approximately 0.36dB per cycle. The user perceives a smooth, gradual increase in low-frequency intensity, rather than a sudden popping sound.

[0071] S504: Apply loudness compensation to the audio signal at a preset audio signal link location, including: Insert an adaptive loudness compensation filter, constructed based on digital filter coefficients, into the audio processing pipeline; The audio processing pipeline is executed in the following order: input PCM signal, room acoustic calibration filtering, user equalizer processing, adaptive loudness compensation filtering, digital volume attenuation control, and output PCM signal.

[0072] For example, if a user simultaneously enables Room Correction (applying -3dB suppression to frequencies below 80Hz) and the user's parametric equalizer (boosting by +2dB at 200Hz), and the current α_smooth = 0.5, then the total gain at 100Hz is calculated as: -3dB + 0dB + 6.1dB - 6dB (volume attenuation) = -2.9dB. Compared to the total gain of -9dB at this frequency without loudness compensation, the perceived low-frequency intensity is increased by 6.1dB, effectively compensating for low-frequency loss at low volumes.

[0073] S505: Also includes a linkage mechanism with scene presets: In response to the user's scene preset switching command, at least one of the upper threshold, lower threshold and maximum gain upper limit parameters in the preset threshold range is updated synchronously; The updated parameters are subjected to direction-aware dual-time-constant adaptive smoothing filtering to avoid timbre jumps during mode switching.

[0074] S6: Perform clipping protection verification on the compensated audio signal and output the audio signal after successful verification.

[0075] This step is the safety assurance and final delivery stage of loudness compensation. It aims to ensure that the output signal amplitude is always kept within the digital full-scale safety limit through forward-looking peak prediction and multi-level protection strategies.

[0076] S601: Clipping protection verification includes a look-ahead peak prediction mechanism: The maximum sample value of the current audio frame is obtained, and the maximum gain across the entire frequency band corresponding to the current target compensation gain curve is determined. Specifically, in this embodiment, the maximum sample value of the current audio frame is obtained in units of audio processing blocks (e.g., 64 sampling points). (Normalized magnitude, value range [0, 1]), from the current target compensation gain curve Extracting the maximum positive gain value across the entire frequency band (Unit: dB); Multiply the maximum sampled value by the linear gain factor corresponding to the maximum gain across the entire frequency band to estimate the maximum signal amplitude that may be achieved after compensation. : ; The system determines whether the maximum signal amplitude exceeds a preset safety threshold. If it does, a clipping protection response is triggered. Specifically, in this embodiment, a safety margin is introduced, and the clipping judgment threshold is set to 0.95 (approximately -0.45 dBFS). This triggers the clipping protection response process.

[0077] Example, current audio frame (Approximately -1.4 dBFS), current loudness compensation (Low-frequency peak gain when α=0.25), calculated as follows The system detected a clipping risk and immediately triggered a protection process.

[0078] S602: The clipping protection response process includes a three-level clipping protection strategy, including: Level 1 Protection: Automatically reduces the global pre-gain to provide margin for loudness compensation. Specifically, in this embodiment, the global pre-gain value applied to the front end of the signal link is reduced by a preset step (e.g., 0.5dB) to provide margin for loudness compensation. This operation does not affect the shape of the compensation curve and has minimal impact on the perceived timbre. The global pre-gain is updated and then recalculated. If the value drops below 0.95, protection is complete; if it still exceeds the limit, proceed to level two protection. Level 2 protection: If the maximum signal amplitude still exceeds the limit after implementing Level 1 protection, the dynamic compensation ratio coefficient is reduced proportionally until the maximum signal amplitude does not exceed the safety threshold. Specifically, in this embodiment, a safety factor is calculated. .use Replace the original This triggers the recalculation of the target curve and the smoothing update of the filter coefficients. The overall compensation intensity is reduced, and the clipping risk decreases accordingly. If the limit is still exceeded after one reduction, iteration continues until the safety conditions are met; Level 3 protection: If the dynamic compensation ratio is reduced to a preset lower limit and the clipping risk still exists, loudness compensation is temporarily and completely disabled, and restored after the preset safety release condition is met. Specifically, in this embodiment, if If the loudness is repeatedly reduced to a preset lower limit (e.g., below 0.1) and the clipping risk still exists, the loudness compensation module will be temporarily and completely disabled (bypassing the compensation filter), and a clipping alarm log will be recorded, including the trigger timestamp, audio content type, and original audio data. Value and final convergence Value. When the peak value of a preset number of consecutive audio frames (e.g., 50 frames) is reached. When all values ​​are below the safety release threshold (e.g., 0.5), it is determined that the high-risk content has passed, compensation is automatically reactivated, and α is restored to the normal calculated value.

[0079] For example, in the above In the scenario, Level 1 protection reduces the global pre-gain by 0.5 dB. It still exceeds the limit. Level 2 protection calculation. , It dropped to about 2.5 dB. It still exceeds the limit. Continue to reduce to , , This precisely meets the safety requirements. The system maintains operation with α=0.08, completely avoiding clipping risks while retaining a small amount of low-frequency compensation.

[0080] S603: Also includes an adaptive learning step for clipping protection events: Record the event information for each trigger of the clipping protection response process and upload it to the cloud data platform and store it in the event log. The event information includes the trigger timestamp, audio content type and dynamic compensation ratio coefficient of final convergence. By analyzing the event distribution patterns in the event log within a preset period, if the frequency of a specific audio input source triggering protection at a preset level or higher exceeds the alarm threshold, the upper limit threshold corresponding to that specific audio input source is automatically lowered to achieve personalized optimization. Specifically, in this embodiment, the event distribution patterns in the event log within a preset period (e.g., 30 days) are analyzed. If the frequency of a specific audio input source triggering protection at level two or higher exceeds the alarm threshold (e.g., 50 times per month), the upper limit threshold corresponding to that specific audio input source is automatically lowered. (For example, reducing the SPL from 75dB to 70dB), fundamentally reducing the aggressiveness of compensation in this scenario. This process requires no user intervention, achieving personalized adaptive evolution that "becomes more and more in line with individual listening habits the more it is used."

[0081] S604: Verified secure audio frames are written to the hardware output buffer, which drives the speakers via the DAC and power amplifier, protecting the state machine from continuous operation.

[0082] After adopting the method of this embodiment, within the full volume range of 30dB to 75dB SPL, the user-perceived frequency response deviation is reduced from an average of ±12 dB to within ±2.5 dB compared to the uncompensated state, with significant improvements in low-frequency fullness and high-frequency airiness. No manual user intervention is required throughout the compensation process; the compensation curve transitions smoothly within 200 ms when the volume changes, with imperceptible timbre shifts. A three-level clipping protection mechanism ensures zero clipping events across the full volume range, and the estimated peak over-limit rate is 0% in engineering verification tests. The loudness compensation module works seamlessly with existing audio processing modules such as room acoustic calibration, user equalizer, and scene presets. In terms of computational efficiency, the 6-section Biquad implementation based on a pre-calculated lookup table has a CPU utilization of only about 0.3%, having a negligible impact on the real-time performance of the embedded platform.

[0083] Second Embodiment Please see Figure 2As shown, based on the same concept, the present invention also provides an adaptive equal loudness curve dynamic loudness compensation system, including an equal loudness curve database preprocessing module, a real-time volume sensing and sound pressure estimation module, an equal loudness difference compensation curve calculation module, a dynamic compensation gain generation module, a filter coefficient conversion and signal link superposition module, and a clipping protection and safe output module. The equal loudness curve database preprocessing module is used to obtain a pre-established high-density equal loudness curve matrix and determine the reference equal loudness curve; The real-time volume sensing and sound pressure estimation module is used to collect the current playback volume of the audio device in real time and calculate the estimated sound pressure level corresponding to the current playback scene. The equal loudness difference compensation curve calculation module is used to interpolate the current equal loudness curve from the high-density equal loudness curve matrix based on the estimated sound pressure level, and to calculate the difference curve between the current equal loudness curve and the reference equal loudness curve. The dynamic compensation gain generation module is used to calculate the dynamic compensation ratio coefficient based on the estimated sound pressure level and the preset threshold range, and to generate the target compensation gain curve by combining the difference curve. The filter coefficient conversion and signal link superposition module is used to convert the target compensation gain curve into digital filter coefficients and to perform loudness compensation on the audio signal at the preset audio signal link position. The clipping protection and safety output module is used to perform clipping protection verification on the compensated audio signal and output the audio signal after successful verification.

[0084] Third Embodiment In this embodiment, a computer device is provided, including a memory and one or more processors. The memory stores computer code. When the computer code is executed by one or more processors, it causes the one or more processors to perform the steps of the adaptive equal-loudness curve dynamic loudness compensation method in the first embodiment.

[0085] In some embodiments of this application, a computer-readable storage medium is also provided, wherein when the computer-readable instructions are executed by one or more processors, the one or more processors perform the steps of an adaptive equal-loudness curve dynamic loudness compensation method as described in any one of the first embodiments.

[0086] The computer device in this embodiment will be described in detail below from the perspective of hardware processing.

[0087] Please see Figure 3 As shown, the computer device includes a processor 100 and a memory 101. The memory 101 stores machine-executable instructions that can be executed by the processor 100. The processor 100 executes the machine-executable instructions to implement the above-described adaptive equal-loudness curve dynamic loudness compensation method.

[0088] further, Figure 3 The computer device shown also includes a bus 102 and a communication interface 103, with the processor 100, communication interface 103 and memory 101 connected via the bus 102.

[0089] The memory 101 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 103 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 102 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0090] The processor 100 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 100 or by instructions in software form. The processor 100 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 101, and the processor 100 reads the information in memory 101 and, in conjunction with its hardware, completes the method steps of the aforementioned embodiments.

[0091] It is understood that, for the aforementioned adaptive equal-loudness curve dynamic loudness compensation method, if all of it is implemented as software functional modules and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer server or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: USB flash drive, mobile hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and other media capable of storing program code.

[0092] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0093] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An adaptive equal-loudness curve dynamic loudness compensation method, characterized in that, Includes the following steps: Obtain a pre-established high-density equal-loudness curve matrix and determine the reference equal-loudness curve; Real-time acquisition of the current playback volume of the audio device, and calculation of the estimated sound pressure level corresponding to the current playback scene; The current equal loudness curve is obtained by interpolation from the high-density equal loudness curve matrix based on the estimated sound pressure level, and the difference curve between the current equal loudness curve and the reference equal loudness curve is calculated. The dynamic compensation ratio coefficient is calculated based on the estimated sound pressure level and the preset threshold range, and the target compensation gain curve is generated by combining the difference curve. The target compensation gain curve is converted into digital filter coefficients, and loudness compensation is performed on the audio signal at a preset audio signal link position. Perform clipping protection verification on the compensated audio signal and output the audio signal after successful verification.

2. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 1, characterized in that, The reference equal loudness curve is the equal loudness curve corresponding to the 83-sound level; The high-density equal-loudness curve matrix is ​​obtained by interpolating and extending the key frequency point data in the ISO 226:2003 standard using cubic spline interpolation, resulting in high-density data covering the range of 20Hz to 20kHz with a resolution of 1 / 12 octave band.

3. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 1, characterized in that, The dynamic compensation ratio coefficient is calculated based on a preset threshold range, including: The current digital volume control value is read from the audio driver layer at a preset sampling period, and the estimated sound pressure level is calculated in combination with the full-power output sound pressure level of the device; Set an upper threshold and a lower threshold, and calculate the initial compensation ratio coefficient within the linear interpolation interval formed by the upper threshold and the lower threshold; When the estimated sound pressure level is greater than or equal to the upper limit threshold, the initial compensation ratio coefficient is forced to be zero; when the estimated sound pressure level is less than or equal to the lower limit threshold, the initial compensation ratio coefficient is forced to be the full value.

4. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 3, characterized in that, The calculation of the dynamic compensation scaling factor is followed by a smoothing filtering mechanism: The initial compensation scaling factor is subjected to a direction-aware dual-time-constant adaptive smoothing filter to obtain the dynamic compensation scaling factor. The smoothing filter process dynamically selects an effective time constant based on the direction of change of the current initial compensation ratio coefficient relative to the previous smoothing coefficient. This allows the dynamic compensation ratio coefficient to adopt a differentiated transition rate in both volume decrease and volume increase scenarios. In this way, it suppresses perceptible timbre jumps caused by abrupt changes in the compensation curve, while improving the timeliness of low-volume compensation establishment and the auditory naturalness of compensation exit when the volume rises.

5. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 1, characterized in that, Calculating the difference curve between the current equal loudness curve and the reference equal loudness curve, and generating a target compensation gain curve by combining the difference curve, including: Calculate the sound pressure level difference between the reference equal loudness curve and the current equal loudness curve at each frequency point to obtain the theoretical compensation requirement curve; The theoretical compensation demand curve is partitioned by frequency band, retaining the complete compensation amount of the low frequency band, linearly reducing the compensation amount of the mid frequency band to zero, and restoring the compensation amount of the preset proportion in the high frequency band to obtain the difference curve after frequency band weighting. The target compensation gain curve is generated by multiplying the frequency band weighted difference curve with the dynamic compensation ratio coefficient.

6. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 1, characterized in that, The target compensation gain curve is converted into digital filter coefficients, including: A pre-calculated lookup table for the dynamic compensation ratio coefficient and the IIR filter coefficient group is pre-established. The IIR filter coefficient group contains several cascaded Biquad sections, which are used for low-frequency low-profile filtering, mid-frequency peak filtering and high-frequency high-profile filtering, respectively. The pre-calculated lookup table is queried based on the current dynamic compensation ratio coefficient, and the target IIR filter coefficient group is obtained through linear interpolation.

7. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 6, characterized in that, After obtaining the target IIR filter coefficient set, a coefficient smoothing update mechanism is also included: The coefficients of the currently running filter are gradually updated using linear interpolation in the coefficient domain; Within each preset update cycle, the currently running filter coefficients are moved closer to the target IIR filter coefficient group by a preset proportional step size until they converge to the target value, in order to eliminate the zipper noise caused by coefficient jumps.

8. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 1, characterized in that, Apply loudness compensation to the audio signal at a preset audio signal link location, including: An adaptive loudness compensation filter, constructed based on the digital filter coefficients, is inserted into the audio processing pipeline; The audio processing pipeline is executed in the following order: input PCM signal, indoor acoustic calibration filtering, user equalizer processing, adaptive loudness compensation filtering, digital volume attenuation control, and output PCM signal.

9. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 4, characterized in that, It also includes a linkage mechanism with preset scenes: In response to the user's scene preset switching command, at least one of the upper limit threshold, lower limit threshold and maximum gain upper limit parameters in the preset threshold range is updated synchronously; The updated parameters are subjected to the aforementioned direction-aware dual-time-constant adaptive smoothing filter to avoid timbre abrupt changes during mode switching.

10. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 1, characterized in that, Clipping protection verification includes a look-ahead peak prediction mechanism: Obtain the maximum sample value of the current audio frame and determine the maximum full-band gain corresponding to the current target compensation gain curve; Multiply the maximum sampled value by the linear gain factor corresponding to the maximum gain across the full bandwidth to estimate the maximum signal amplitude that may be achieved after compensation; Determine whether the maximum signal amplitude is greater than a preset safety threshold. If it is, trigger the clipping protection response process.

11. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 10, characterized in that, The clipping protection response process includes a three-level clipping protection strategy, including: Level 1 protection: Automatically reduces global pregain to provide margin for loudness compensation; Level 2 protection: If the maximum signal amplitude still exceeds the limit after implementing Level 1 protection, the dynamic compensation ratio coefficient is reduced proportionally until the maximum signal amplitude is not greater than the safety threshold. Level 3 protection: If the dynamic compensation ratio coefficient is reduced to a preset lower limit and the clipping risk still exists, loudness compensation is temporarily and completely disabled, and then restored after the preset safety release conditions are met.

12. The adaptive equal-loudness curve dynamic loudness compensation method according to claim 10, characterized in that, It also includes an adaptive learning step for clipping protection events: Record the event information for each trigger of the clipping protection response process and store it in the event log. The event information includes the trigger timestamp, audio content type, and dynamic compensation ratio coefficient for final convergence. Analyze the event distribution patterns in the event log within a preset period. If the frequency of a specific audio input source triggering protection at a preset level or higher exceeds the alarm threshold, the upper limit threshold corresponding to that specific audio input source will be automatically lowered to achieve personalized optimization.

13. An adaptive equal-loudness curve dynamic loudness compensation system, characterized in that, It includes a preprocessing module for equal loudness curve database, a real-time volume sensing and sound pressure estimation module, an equal loudness difference compensation curve calculation module, a dynamic compensation gain generation module, a filter coefficient conversion and signal link superposition module, and a clipping protection and safe output module. The equal loudness curve database preprocessing module is used to obtain a pre-established high-density equal loudness curve matrix and determine a reference equal loudness curve; The real-time volume sensing and sound pressure estimation module is used to collect the current playback volume of the audio device in real time and calculate the estimated sound pressure level corresponding to the current playback scene. The equal loudness difference compensation curve calculation module is used to interpolate the current equal loudness curve from the high-density equal loudness curve matrix based on the estimated sound pressure level, and to calculate the difference curve between the current equal loudness curve and the reference equal loudness curve. The dynamic compensation gain generation module is used to calculate the dynamic compensation ratio coefficient based on the estimated sound pressure level and the preset threshold range, and generate the target compensation gain curve in combination with the difference curve. The filter coefficient conversion and signal link superposition module is used to convert the target compensation gain curve into digital filter coefficients and perform loudness compensation on the audio signal at a preset audio signal link position. The clipping protection and safety output module is used to perform clipping protection verification on the compensated audio signal and output the audio signal after successful verification.

14. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the adaptive equal-loudness curve dynamic loudness compensation method as described in any one of claims 1-12.

15. A computer-readable storage medium, characterized in that, When the computer-readable instructions are executed by one or more processors, the one or more processors perform the steps of the adaptive equal-loudness curve dynamic loudness compensation method as described in any one of claims 1 to 12.

Images