Hybrid high-precision frequency weighting method and system and electronic equipment

Abstract

The invention discloses a hybrid high-precision frequency weighting method and system and electronic equipment, and the method comprises the steps: obtaining an original sound signal, carrying out the analog-to-digital conversion, and carrying out the preprocessing of the sound signal after the analog-to-digital conversion, so as to obtain a signal frequency spectrum; segmenting the signal spectrum, and weighting each segment to obtain a first weighting sound level and a second weighting sound level; superposing the first weighted sound level and the second weighted sound level to determine a total weighted sound level; and performing dynamic error calibration on the total weighted sound level, determining a standard error, and when the standard error is smaller than or equal to a preset error threshold value, outputting the total weighted sound level. According to the invention, a differentiated optimization strategy is adopted, and the attenuation deviation of a traditional digital filter in a low-frequency area is effectively corrected in a low-frequency band; in the middle and high frequency bands, the accuracy is ensured, meanwhile, the occupation of operation resources is remarkably reduced, the signal processing delay time is shortened, and the response requirement of real-time noise monitoring can be met.

Description

Technical Field

[0001] This invention relates to the field of noise measurement technology, and more specifically, to a hybrid high-precision frequency weighting method, system, and electronic device. Background Technology

[0002] Frequency weighting is a core component of noise measurement, aiming to simulate the sensitivity characteristics of the human ear to sounds at different frequencies. Traditional noise measurement instruments use analog circuits composed of resistors, capacitors, and integrated operational amplifiers to achieve frequency weighting, which has inherent drawbacks such as weak anti-interference capability, the ability to select only a single weighting network in a single measurement, and the inability to simultaneously measure multiple weighted sound levels.

[0003] With the development of digital signal processing technology, digital frequency weighting has become a mainstream technology, mainly divided into two categories: time-domain filtering based on digital filters and frequency-domain calculation based on Fast Fourier Transform (FFT). Time-domain filtering often uses Infinite Impulse Response (IIR) digital filters, digitizing analog filters through bilinear transform. It has the advantages of simple implementation and good real-time performance, but in the low-frequency range, it suffers from significant amplitude-frequency characteristic distortion due to the inherent nonlinear distortion of the bilinear transform, and the quantization error of the filter coefficients further affects accuracy. Frequency-domain calculation uses FFT to directly perform weighting and attenuation calculations in the frequency domain, improving the accuracy of low-frequency weighting. However, it has a large computational load and high memory consumption, making it difficult to meet high real-time requirements. Furthermore, it is also susceptible to spectral leakage and picket-fence effects, which introduce measurement errors.

[0004] Therefore, existing single-mode digital solutions face a technical contradiction in balancing high precision across the entire frequency band (especially low frequencies) with high real-time performance, failing to meet the stringent requirements of high-end noise measurement instruments for both accuracy and real-time performance. A hybrid high-precision frequency weighting method is needed. Summary of the Invention

[0005] This invention proposes a hybrid high-precision frequency weighting method, system, and electronic device to solve the problem of how to achieve high-precision frequency weighting.

[0006] To address the aforementioned problems, according to one aspect of the present invention, a hybrid high-precision frequency weighting method is provided, the method comprising:

[0007] The original acoustic signal is acquired, analog-to-digital conversion is performed on the original acoustic signal, and the converted acoustic signal is preprocessed to obtain the signal spectrum;

[0008] The signal spectrum is segmented, and each segment is weighted to obtain a first weighted sound level and a second weighted sound level.

[0009] The first weighted sound level and the second weighted sound level are superimposed to determine the total weighted sound level;

[0010] The total weighted sound level is dynamically calibrated to determine the standard error, and the total weighted sound level is output when the standard error is less than or equal to a preset error threshold.

[0011] Preferably, the preprocessing of the analog-to-digital converted acoustic signal to obtain the signal spectrum includes:

[0012] Anti-aliasing filtering is applied to the acoustic signal after analog-to-digital conversion.

[0013] The signal after anti-aliasing filtering is framed, a short-time Fourier transform is performed on each frame, and the proportion of low-frequency components within the frame is calculated.

[0014] The window function is determined based on the proportion of low-frequency components within the frame.

[0015] The window function is applied to the signal frame, and a short-time Fourier transform is performed to obtain the signal spectrum. Preferably, the method calculates the proportion η of low-frequency components within the frame using the following method:

[0016] η = (Total sound energy in the 20Hz to 1kHz frequency band) / (Total sound energy in the entire 20Hz to 20kHz frequency band)

[0017] Wherein, using E(f)=|X(f)| 2 Calculate the sound energy, where f is the frequency of the sound signal and X(f) is the result of the short-time Fourier transform.

[0018] Preferably, determining the window function based on the proportion of low-frequency components within the frame includes:

[0019] If the proportion of low-frequency components within the frame is η≥30%, then the Blackman window is selected as the window function; if the proportion of low-frequency components within the frame is η<30%, then the Hanning window is selected as the window function.

[0020] Preferably, the method of segmenting the signal spectrum and weighting each segment to obtain a first weighted sound level and a second weighted sound level includes:

[0021] The signal spectrum is divided into two segments based on a dividing frequency of 1kHz.

[0022] For low-frequency signal spectra below 1kHz, a modified fast Fourier transform method is used for weighting to obtain the first weighted sound level; for mid-to-high frequency signal spectra of 1kHz and above, an optimized infinite impulse response digital filter method is used for weighting to obtain the second weighted sound level.

[0023] Preferably, the weighting is performed using a modified Fast Fourier Transform method to obtain the first weighted sound level, including:

[0024] Zero-padding is applied to the spectrum of low-frequency signals.

[0025] For any frequency point f located between adjacent spectral lines, use its left and right adjacent spectral lines f k f k+1 Weighted sound level L k L k+1 and its corresponding sound energy E k E k+1 The acoustic energy E(f) at point f is calculated using cubic spline interpolation.

[0026] The acoustic energy E(f) at each frequency point f is converted into the corresponding weighted sound level to obtain the first weighted sound level.

[0027] Preferably, the weighting is performed using an optimized infinite impulse response digital filter method to obtain the second weighted sound level, including:

[0028] An infinite impulse response digital filter quantized with 32-bit floating-point coefficients is used to perform real-time recursive filtering on the spectrum of mid-to-high frequency signals to obtain a weighted time-domain signal.

[0029] The short-time power spectrum of the time-domain signal is calculated and converted into acoustic energy.

[0030] The sound energy at each frequency point is converted into the corresponding weighted sound level to obtain the second weighted sound level;

[0031] Specifically, the pre-distortion formula in the bilinear transform is modified by introducing a pole fine-tuning coefficient k, thus revising the pre-distortion formula to Ω = 2kf. s tan(πf / f s This is to compensate for the amplitude-frequency response deviation caused by the bilinear transform; where Ω is the filter angular frequency; k is the pole fine-tuning coefficient; fs is the acoustic signal sampling frequency; and f is the acoustic signal frequency.

[0032] An infinite impulse response digital filter is implemented by cascading multiple second-order direct type II networks. The A-weighted filter uses three cascaded second-order direct type II networks, with the transfer function being... The C-weighted filter uses two cascaded second-order direct type II networks, with the transfer function being... Perform 32-bit floating-point quantization on the filter coefficients; where c and d represent the relevant undetermined coefficients, and z is the transform sign.

[0033] Preferably, the method of superimposing the first weighted sound level and the second weighted sound level to determine the total weighted sound level includes:

[0034] L total =10lg(E low +E high),

[0035]

[0036] Among them, L total For total weighted sound level; L low (f) represents the first weighted sound level; L high (f) represents the second weighted sound level; E low E represents the total acoustic energy in the low-frequency range. high This represents the total acoustic energy in the high-frequency band.

[0037] Preferably, the dynamic error calibration of the total weighted sound level to determine the standard error includes:

[0038] Select key frequency points, output sinusoidal signals with a peak value of 1V and a frequency corresponding to each selected key frequency point through a standard signal generator, and calculate the weighted output value L. meas (f) Based on the weighted output value and the standard weighted value L std (f) Calculate the calibration coefficient δ(f) = L std (f)-L meas (f), and establish a table showing the correspondence between the calibration coefficient δ and the frequency f;

[0039] For each frequency point f in the signal to be weighted, the corresponding calibration coefficient δ(f) is obtained from the calibration coefficient-frequency correspondence table through linear interpolation, and the uncalibrated weighted sound level L is then... raw (f) Make corrections to obtain the calibrated sound level L. cal (f)=L raw (f)+δ(f);

[0040] A standard signal with a frequency of 1kHz and a peak-to-peak value of 1V is generated using a standard signal generator. Single-point frequency verification is performed to determine the calibration error ΔL = L. raw (f)+|L cal (1kHz)-0|.

[0041] Preferably, the method further includes:

[0042] If the calibration error is greater than the preset error threshold, the calibration coefficients of all key frequency points are remeasured and the correspondence table between the calibration coefficient δ and the frequency f is updated.

[0043] According to another aspect of the present invention, a hybrid high-precision frequency weighting system is provided, the system comprising:

[0044] The signal processing unit is used to acquire the original acoustic signal, perform analog-to-digital conversion on the original acoustic signal, and preprocess the analog-to-digital converted acoustic signal to obtain the signal spectrum;

[0045] The segmented weighting unit is used to segment the signal spectrum and perform weighting on each segment to obtain a first weighted sound level and a second weighted sound level.

[0046] A sound level superposition unit is used to superimpose the first weighted sound level and the second weighted sound level to determine the total weighted sound level;

[0047] The dynamic error calibration unit is used to perform dynamic error calibration on the total weighted sound level, determine the standard error, and output the total weighted sound level when the standard error is less than or equal to a preset error threshold.

[0048] Preferably, the signal processing unit preprocesses the analog-to-digital converted acoustic signal to obtain the signal spectrum, including:

[0049] Anti-aliasing filtering is applied to the acoustic signal after analog-to-digital conversion.

[0050] The signal after anti-aliasing filtering is framed, a short-time Fourier transform is performed on each frame, and the proportion of low-frequency components within the frame is calculated.

[0051] The window function is determined based on the proportion of low-frequency components within the frame.

[0052] The window function is applied to the signal frame, and a short-time Fourier transform is performed to obtain the signal spectrum.

[0053] Preferably, the signal processing unit calculates the proportion η of intra-frame low-frequency components using the following method:

[0054] η = (Total sound energy in the 20Hz to 1kHz frequency band) / (Total sound energy in the entire 20Hz to 20kHz frequency band)

[0055] Wherein, using E(f)=|X(f)| 2 Calculate the sound energy, where f is the frequency of the sound signal and X(f) is the result of the short-time Fourier transform.

[0056] Preferably, the signal processing unit determines the window function based on the proportion of low-frequency components within the frame, including:

[0057] If the proportion of low-frequency components within the frame is η≥30%, then the Blackman window is selected as the window function; if the proportion of low-frequency components within the frame is η<30%, then the Hanning window is selected as the window function.

[0058] Preferably, the segmented weighting unit segments the signal spectrum and weights each segment to obtain a first weighted sound level and a second weighted sound level, comprising:

[0059] The signal spectrum is divided into two segments based on a dividing frequency of 1kHz.

[0060] For low-frequency signal spectra below 1kHz, a modified fast Fourier transform method is used for weighting to obtain the first weighted sound level; for mid-to-high frequency signal spectra of 1kHz and above, an optimized infinite impulse response digital filter method is used for weighting to obtain the second weighted sound level.

[0061] Preferably, the segmented weighting unit uses a modified fast Fourier transform method for weighting to obtain a first weighted sound level, including:

[0062] Zero-padding is applied to the spectrum of low-frequency signals.

[0063] For any frequency point f located between adjacent spectral lines, use its left and right adjacent spectral lines f k f k+1 Weighted sound level L k L k+1 and its corresponding sound energy E k E k+1 The acoustic energy E(f) at point f is calculated using cubic spline interpolation.

[0064] The acoustic energy E(f) at each frequency point f is converted into the corresponding weighted sound level to obtain the first weighted sound level.

[0065] Preferably, the segmented weighting unit employs an optimized infinite impulse response digital filter method for weighting to obtain a second weighted sound level, comprising:

[0066] An infinite impulse response digital filter quantized with 32-bit floating-point coefficients is used to perform real-time recursive filtering on the spectrum of mid-to-high frequency signals to obtain a weighted time-domain signal.

[0067] The short-time power spectrum of the time-domain signal is calculated and converted into acoustic energy.

[0068] The sound energy at each frequency point is converted into the corresponding weighted sound level to obtain the second weighted sound level;

[0069] Specifically, the pre-distortion formula in the bilinear transform is modified by introducing a pole fine-tuning coefficient k, thus revising the pre-distortion formula to Ω = 2kf. s tan(πf / f s This is to compensate for the amplitude-frequency response deviation caused by the bilinear transform; where Ω is the filter angular frequency; k is the pole fine-tuning coefficient; fs is the acoustic signal sampling frequency; and f is the acoustic signal frequency.

[0070] An infinite impulse response digital filter is implemented by cascading multiple second-order direct type II networks. The A-weighted filter uses three cascaded second-order direct type II networks, with the transfer function being... The C-weighted filter uses two cascaded second-order direct type II networks, with the transfer function being... Perform 32-bit floating-point quantization on the filter coefficients; where c and d represent the relevant undetermined coefficients, and z is the transform sign.

[0071] Preferably, the sound level superposition unit superimposes the first weighted sound level and the second weighted sound level to determine the total weighted sound level, including:

[0072] L total =10lg(E low +E high ),

[0073]

[0074] Among them, L total For total weighted sound level; L low (f) represents the first weighted sound level; L high (f) represents the second weighted sound level; E low E represents the total acoustic energy in the low-frequency range. high This represents the total acoustic energy in the high-frequency band.

[0075] Preferably, the dynamic calibration unit performs dynamic error calibration on the total weighted sound level to determine the standard error, including:

[0076] Select key frequency points, output sinusoidal signals with a peak value of 1V and a frequency corresponding to each selected key frequency point through a standard signal generator, and calculate the weighted output value L. meas (f) Based on the weighted output value and the standard weighted value L std (f) Calculate the calibration coefficient δ(f) = L std (f)-L meas (f), and establish a table showing the correspondence between the calibration coefficient δ and the frequency f;

[0077] For each frequency point f in the signal to be weighted, the corresponding calibration coefficient δ(f) is obtained from the calibration coefficient-frequency correspondence table through linear interpolation, and the uncalibrated weighted sound level L is then... raw (f) Make corrections to obtain the calibrated sound level L. cal (f)=L raw (f)+δ(f);

[0078] A standard signal with a frequency of 1kHz and a peak-to-peak value of 1V is generated using a standard signal generator. Single-point frequency verification is performed to determine the calibration error ΔL = L. raw (f)+|L cal (1kHz)-0|.

[0079] Preferably, the dynamic calibration unit is further configured to:

[0080] If the calibration error is greater than the preset error threshold, the calibration coefficients of all key frequency points are remeasured and the correspondence table between the calibration coefficient δ and the frequency f is updated.

[0081] According to another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of a hybrid high-precision frequency weighting method.

[0082] According to another aspect of the present invention, the present invention provides an electronic device, comprising:

[0083] The aforementioned computer-readable storage medium; and

[0084] One or more processors for executing a program in the computer-readable storage medium.

[0085] This invention provides a hybrid high-precision frequency weighting method, system, and electronic device, comprising: acquiring an original acoustic signal; performing analog-to-digital conversion on the original acoustic signal; and preprocessing the converted acoustic signal to obtain a signal spectrum; segmenting the signal spectrum and weighting each segment to obtain a first weighted sound level and a second weighted sound level; superimposing the first weighted sound level and the second weighted sound level to determine a total weighted sound level; performing dynamic error calibration on the total weighted sound level to determine a standard error; and outputting the total weighted sound level when the standard error is less than or equal to a preset error threshold. This invention employs differentiated optimization strategies for different frequency bands of noise signals. In the low-frequency band, an improved Fast Fourier Transform combined with zero-padding and spectral interpolation effectively corrects the attenuation deviation of traditional digital filters in the low-frequency region. In the mid-to-high frequency band, a simplified infinite impulse response digital filter is used, which, while ensuring accuracy, significantly reduces computational resource consumption and shortens signal processing delay time compared to the pure Fast Fourier Transform method, thus meeting the response requirements of real-time noise monitoring. Furthermore, by introducing adaptive window functions, pole fine-tuning, and dynamic error calibration mechanisms, the influence of factors such as spectrum leakage and quantization errors can be effectively suppressed, ensuring the stability and reliability of frequency weighting results across the entire frequency band. Attached Figure Description

[0086] Exemplary embodiments of the present invention can be more fully understood by referring to the following figures:

[0087] Figure 1 This is a flowchart of a hybrid high-precision frequency weighting method 100 according to an embodiment of the present invention;

[0088] Figure 2 This is a flowchart of a hybrid high-precision frequency weighting process according to an embodiment of the present invention;

[0089] Figure 3 This is a schematic diagram of the hardware device for implementing hybrid high-precision frequency weighting according to an embodiment of the present invention.

[0090] Figure 4 This is a schematic diagram of the structure of a hybrid high-precision frequency weighting system 400 according to an embodiment of the present invention. Detailed Implementation

[0091] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0092] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0093] Figure 1 This is a flowchart of a hybrid high-precision frequency weighting method 100 according to an embodiment of the present invention. Figure 1 As shown, the hybrid high-precision frequency weighting method provided by this invention employs differentiated optimization strategies for different frequency bands of the noise signal. In the low-frequency band, an improved Fast Fourier Transform combined with zero-padding expansion and spectral interpolation effectively corrects the attenuation deviation of traditional digital filters in the low-frequency region. In the mid-to-high frequency band, a simplified infinite impulse response digital filter is used, which significantly reduces computational resource consumption and shortens signal processing delay time compared to the pure Fast Fourier Transform method while ensuring accuracy, thus meeting the response requirements of real-time noise monitoring. Furthermore, by introducing adaptive window functions, pole fine-tuning, and dynamic error calibration mechanisms, the influence of factors such as spectral leakage and quantization errors can be effectively suppressed, ensuring the stability and reliability of the frequency weighting results across the entire frequency band. The hybrid high-precision frequency weighting method 100 provided by this invention begins at step 101. In step 101, the original acoustic signal is acquired, the original acoustic signal is converted from analog to digital, and the converted acoustic signal is preprocessed to obtain the signal spectrum.

[0094] Preferably, the preprocessing of the analog-to-digital converted acoustic signal to obtain the signal spectrum includes:

[0095] Anti-aliasing filtering is applied to the acoustic signal after analog-to-digital conversion.

[0096] The signal after anti-aliasing filtering is framed, a short-time Fourier transform is performed on each frame, and the proportion of low-frequency components within the frame is calculated.

[0097] The window function is determined based on the proportion of low-frequency components within the frame.

[0098] The window function is applied to the signal frame, and a short-time Fourier transform is performed to obtain the signal spectrum. Preferably, the method calculates the proportion η of low-frequency components within the frame using the following method:

[0099] η = (Total sound energy in the 20Hz to 1kHz frequency band) / (Total sound energy in the entire 20Hz to 20kHz frequency band)

[0100] Wherein, using E(f)=|X(f)| 2 Calculate the sound energy, where f is the frequency of the sound signal and X(f) is the result of the short-time Fourier transform.

[0101] Preferably, determining the window function based on the proportion of low-frequency components within the frame includes:

[0102] If the proportion of low-frequency components within the frame is η≥30%, then the Blackman window is selected as the window function; if the proportion of low-frequency components within the frame is η<30%, then the Hanning window is selected as the window function.

[0103] The implementation of the method of the present invention includes four stages executed sequentially: acoustic signal preprocessing, frequency segmentation weighting, segmentation result energy synthesis, and dynamic error calibration.

[0104] Combination Figure 2 As shown in the invention, the acquired acoustic signal first needs to be preprocessed. The purpose of the acoustic signal preprocessing stage is to perform anti-aliasing filtering on the original acoustic signal after analog-to-digital conversion, and to adaptively select a window function based on the signal spectral characteristics to suppress spectral aliasing and spectral leakage, thereby obtaining a high-fidelity frequency domain signal.

[0105] Specifically, the acoustic signal preprocessing includes the following steps:

[0106] First, an 8th-order elliptic low-pass filter is used to perform anti-aliasing filtering on the signal. Its passband is 20Hz to 24kHz, passband ripple ≤0.1dB, stopband attenuation ≥80dB, and filtering delay ≤50μs.

[0107] Secondly, the filtered signal is divided into frames, a short-time Fourier transform is performed on each frame, and the proportion of low-frequency components η within the frame is calculated, where η = (total sound energy in the 20Hz to 1kHz frequency band) / (total sound energy in the 20Hz to 20kHz full frequency band).

[0108] Then, the window function is dynamically selected based on the value of η: if η≥30%, it is determined that the low-frequency component in the original sound signal is dominant, and the Blackman window with better sidelobe attenuation performance is selected to effectively suppress energy crosstalk between adjacent frequencies in the low-frequency band; if η<30%, it is determined that the mid-high frequency component in the original sound signal is dominant, and the Hanning window with a narrower main lobe width is selected to improve the frequency resolution of the mid-high frequency band and reduce the influence of the picket fence effect.

[0109] Finally, after applying the selected window function to the signal frame, a high-resolution short-time Fourier transform is performed to obtain the frequency domain spectrum for frequency segmentation weighting.

[0110] Preferably, the formula for calculating sound energy is E(f)= / X(f) / 2 Where f is the frequency of the acoustic signal, and X(f) is the result of the short-time Fourier transform.

[0111] In step 102, the signal spectrum is segmented and each segment is weighted to obtain a first weighted sound level and a second weighted sound level.

[0112] Preferably, the method of segmenting the signal spectrum and weighting each segment to obtain a first weighted sound level and a second weighted sound level includes:

[0113] The signal spectrum is divided into two segments based on a dividing frequency of 1kHz.

[0114] For low-frequency signal spectra below 1kHz, a modified fast Fourier transform method is used for weighting to obtain the first weighted sound level; for mid-to-high frequency signal spectra of 1kHz and above, an optimized infinite impulse response digital filter method is used for weighting to obtain the second weighted sound level.

[0115] Preferably, the weighting is performed using a modified Fast Fourier Transform method to obtain the first weighted sound level, including:

[0116] Zero-padding is applied to the spectrum of low-frequency signals.

[0117] For any frequency point f located between adjacent spectral lines, use its left and right adjacent spectral lines f k f k+1 Weighted sound level L k L k+1 and its corresponding sound energy E k E k+1 The acoustic energy E(f) at point f is calculated using cubic spline interpolation.

[0118] The acoustic energy E(f) at each frequency point f is converted into the corresponding weighted sound level to obtain the first weighted sound level.

[0119] Preferably, the weighting is performed using an optimized infinite impulse response digital filter method to obtain the second weighted sound level, including:

[0120] An infinite impulse response digital filter quantized with 32-bit floating-point coefficients is used to perform real-time recursive filtering on the spectrum of mid-to-high frequency signals to obtain a weighted time-domain signal.

[0121] The short-time power spectrum of the time-domain signal is calculated and converted into acoustic energy.

[0122] The sound energy at each frequency point is converted into the corresponding weighted sound level to obtain the second weighted sound level;

[0123] Specifically, the pre-distortion formula in the bilinear transform is modified by introducing a pole fine-tuning coefficient k, thus revising the pre-distortion formula to Ω = 2kf. s tan(πf / f s This is to compensate for the amplitude-frequency response deviation caused by the bilinear transform; where Ω is the filter angular frequency; k is the pole fine-tuning coefficient; fs is the acoustic signal sampling frequency; and f is the acoustic signal frequency.

[0124] An infinite impulse response digital filter is implemented by cascading multiple second-order direct type II networks. The A-weighted filter uses three cascaded second-order direct type II networks, with the transfer function being... The C-weighted filter uses two cascaded second-order direct type II networks, with the transfer function being... Perform 32-bit floating-point quantization on the filter coefficients; where c and d represent the relevant undetermined coefficients, and z is the transform sign.

[0125] Combination Figure 2 As shown, in the frequency segmentation and weighting stage of this invention, the preprocessed signal spectrum is segmented with 1kHz as the dividing frequency, and different optimization algorithms are used to process them respectively.

[0126] For the low-frequency band below 1kHz, an improved fast Fourier transform method is used for weighting; for the mid-to-high frequency band above 1kHz, an optimized infinite impulse response digital filter method is used for weighting, thereby achieving a balance between the accuracy and efficiency of frequency weighting calculation.

[0127] Specifically, the improved Fast Fourier Transform method includes the following steps:

[0128] First, the preprocessed signal is zero-padding and extended, expanding the 1024-point Fast Fourier Transform operation to 2048 points, improving the frequency resolution to 23.4375Hz, and solving the weighting error caused by the sparse spectral lines in the low-frequency band.

[0129] Secondly, for any frequency point f located between adjacent spectral lines, its left and right adjacent spectral lines f are used. k f k+1 Weighted sound level L k L k+1 and its corresponding sound energy The acoustic energy E(f) at point f is calculated using cubic spline interpolation: E(f) = a0 + a1(ff) k )+a2(ff k ) 2 +a3(ff k ) 3 The coefficients a0, a1, a2, and a3 are determined by the boundary condition E(f k ) = E k 、E(f k+1 ) = E k+1 And the second derivative is obtained by continuous solution;

[0130] Then, the acoustic energy E(f) at frequency point f is converted into the weighted sound level L(f) = 10lgE(f);

[0131] Finally, based on the A and C weighting formulas specified in the IEC 61672-1 standard, 64-bit floating-point arithmetic is used to calculate the weighted attenuation value for each frequency point to avoid truncation errors, ensure the accuracy of attenuation value calculation, and store it in the Flash memory of the core processing module.

[0132] Specifically, the optimized infinite impulse response digital filter method includes the following steps:

[0133] First, the pre-distortion formula in the bilinear transform is modified by introducing a pole fine-tuning coefficient k, thus revising the pre-distortion formula to Ω = 2kf. s tan(πf / f s This is to compensate for the amplitude-frequency response deviation caused by the bilinear transformation;

[0134] Then, an infinite impulse response digital filter is implemented by cascading multiple second-order direct type II networks. The A-weighted filter uses three cascaded second-order direct type II networks, with the transfer function being... The C-weighted filter uses two cascaded second-order direct type II networks, with the transfer function being... Perform 32-bit floating-point quantization on the filter coefficients;

[0135] Secondly, an optimized IIR filter quantized with 32-bit floating-point coefficients is used to perform real-time recursive filtering on the preprocessed acoustic signal to obtain the weighted time-domain signal.

[0136] Next, the short-time power spectrum of the filtered output time-domain signal is calculated and converted into acoustic energy;

[0137] Finally, the sound energy at each frequency point is converted into the corresponding weighted sound level to obtain the second weighted sound level.

[0138] Preferably, the pole fine-tuning coefficient k ranges from 0.98 to 1.02.

[0139] In step 103, the first weighted sound level and the second weighted sound level are superimposed to determine the total weighted sound level.

[0140] Preferably, the method of superimposing the first weighted sound level and the second weighted sound level to determine the total weighted sound level includes:

[0141] L total =10lg(E low +E high ),

[0142]

[0143] Among them, L total For total weighted sound level; L low (f) represents the first weighted sound level; L high (f) represents the second weighted sound level; E low E represents the total acoustic energy in the low-frequency range. high This represents the total acoustic energy in the high-frequency band.

[0144] Combination Figure 2 As shown, in the present invention, in the energy synthesis stage of the segmented results, the principle of sound level superposition is adopted to synthesize the first weighted sound level of the low frequency band and the second weighted sound level of the mid-high frequency band obtained by frequency segment weighting processing to determine the total weighted sound level.

[0145] Specifically, the calculation process includes:

[0146] Calculate the total acoustic energy in the low-frequency range: Where, f∈[20Hz,1kHz], L low (f) represents the first weighted sound level in the low-frequency band;

[0147] Calculate the total acoustic energy in the mid-to-high frequency range: Where, f∈[1kHz,20kHz], L high (f) represents the second weighted sound level in the mid-to-high frequency band;

[0148] Calculate the total weighted sound level of the synthesized sound: L total =10lg(E low +E high ).

[0149] Preferably, the entire synthesis process uses 64-bit floating-point operations to avoid precision loss during energy superposition.

[0150] In step 104, the total weighted sound level is dynamically calibrated to determine the standard error, and when the standard error is less than or equal to a preset error threshold, the total weighted sound level is output.

[0151] Preferably, the dynamic error calibration of the total weighted sound level to determine the standard error includes:

[0152] Select key frequency points, output sinusoidal signals with a peak value of 1V and a frequency corresponding to each selected key frequency point through a standard signal generator, and calculate the weighted output value L. meas (f) Based on the weighted output value and the standard weighted value L std (f) Calculate the calibration coefficient δ(f) = L std (f)-L meas (f), and establish a table showing the correspondence between the calibration coefficient δ and the frequency f;

[0153] For each frequency point f in the signal to be weighted, the corresponding calibration coefficient δ(f) is obtained from the calibration coefficient-frequency correspondence table through linear interpolation, and the uncalibrated weighted sound level L is then... raw (f) Make corrections to obtain the calibrated sound level L. cal (f)=L raw (f)+δ(f);

[0154] A standard signal with a frequency of 1kHz and a peak-to-peak value of 1V is generated using a standard signal generator. Single-point frequency verification is performed to determine the calibration error ΔL = L. raw (f)+|L cal (1kHz)-0|.

[0155] Preferably, the method further includes:

[0156] If the calibration error is greater than the preset error threshold, the calibration coefficients of all key frequency points are remeasured and the correspondence table between the calibration coefficient δ and the frequency f is updated.

[0157] In this invention, during the dynamic error calibration stage, drift errors that may occur during long-term system operation are corrected to ensure the long-term stability of weighting accuracy.

[0158] Combination Figure 2 As shown, the specific steps include:

[0159] First, 23 key frequency points specified in the IEC 61672-1 standard are selected, including 12.5Hz, 16Hz, 20Hz, ..., 20kHz. A standard signal generator outputs a sinusoidal signal with a peak-to-peak value of 1V and a frequency corresponding to each selected frequency point. The weighted output value L of the hybrid high-precision frequency weighting digital implementation method described in this invention is then measured. meas(f), and compared with the standard weighted value L std (f) is compared, and the calibration coefficient δ(f) = L is calculated. std (f)-L meas (f) Establish a table showing the correspondence between calibration coefficient δ and frequency f, and store it in the Flash memory of the core processing module;

[0160] Then, for each frequency point f in the signal to be weighted, the corresponding calibration coefficient δ(f) is obtained from the calibration coefficient-frequency correspondence table through linear interpolation, and the uncalibrated weighted sound level L is then... raw (f) Make corrections to obtain the calibrated sound level L. cal (f)=L raw (f)+δ(f);

[0161] Finally, the system triggers an internal automatic calibration process every hour, where a standard signal generator produces a standard signal with a frequency of 1kHz and a peak-to-peak value of 1V for single-point frequency verification. If the calibration error ΔL = L raw (f)+|L cal (1kHz) - If 0|>0.5dB, then remeasure the calibration coefficient K at the key frequency point. i And update the table of correspondence between calibration coefficients and frequencies.

[0162] The method of the present invention can be based on Figure 3 The aforementioned hybrid high-precision frequency weighting digitization device includes a signal acquisition module, a core processing module, an output and control module, and a power supply module. Specifically, 201 is an electret microphone, 202 is a preamplifier, 203 is a 24-bit high-precision analog-to-digital converter, 204 is a digital signal processor, 205 is SDRAM, 206 is Flash memory, 207 is an LCD display, and 208 is the power supply module.

[0163] The signal acquisition module is used to convert sound pressure signals into high-precision digital signals. Specifically, it includes: an electret microphone with a frequency response range of 20Hz-20kHz; a preamplifier for amplifying the output signal of the electret microphone to a suitable level; and a 24-bit high-precision analog-to-digital converter that transmits digital audio streams at a rate of not less than 2.304Mbps via an I2S interface.

[0164] The core processing module, serving as the device's computing center, is powered by a 32-bit floating-point digital signal processor with a clock speed of at least 300MHz. This module is equipped with at least 256MB of SDRAM as data cache and program execution space, and at least 64MB of Flash memory for storing programs, filter coefficients, calibration coefficients, and frequency correspondence tables. This module efficiently receives analog-to-digital converted data using direct memory access technology and is responsible for executing all algorithmic steps in the hybrid high-precision frequency weighting digital implementation method.

[0165] The output and control module, used for human-computer interaction and data output, specifically includes: a 128×64 dot matrix LCD display screen for real-time display of A and C weighted sound levels, spectrum, and system status; several physical buttons for triggering functions such as calibration and mode switching; and a data output interface that supports analog output or network data transmission based on the TCP / IP protocol.

[0166] The power module has an input terminal connected to AC power and an output with multi-channel isolation design. The signal acquisition module, core processing module, and output and control module provide corresponding operating voltages.

[0167] The connections between the modules are as follows: In the signal acquisition module, the electret microphone and the preamplifier are electrically connected via wires, and the output of the preamplifier is connected to the analog input of the 24-bit high-precision analog-to-digital converter (ADC). The digital output of the 24-bit ADC is connected to the data input port of the digital signal processor in the core processing module via a high-speed serial audio interface, forming a digital acquisition link for the sound signal. The core processing module, as the central control and data processing unit, has its ADC connected to the physical buttons and status indicator lights in the output control module via general-purpose input / output ports to respond to user commands and indicate system status. Simultaneously, the ADC drives the LCD display for local display via a parallel bus or high-speed serial interface and controls the data output interface for remote data communication via its peripheral interface.

[0168] The method of this invention combines the low-frequency accuracy advantage of frequency domain calculation based on fast Fourier transform with the high-efficiency processing capability of time domain filtering based on infinite impulse response digital filter by using a frequency band optimization processing strategy. It also introduces an adaptive preprocessing and dynamic calibration mechanism, which effectively solves the problem of not being able to balance high accuracy and high real-time performance across the entire frequency band, and significantly improves the long-term stability of the system.

[0169] Figure 4 This is a schematic diagram of the structure of a hybrid high-precision frequency weighting system 400 according to an embodiment of the present invention. Figure 4As shown, the hybrid high-precision frequency weighting system 400 provided in this embodiment of the invention includes: a signal processing unit 401, a segmented weighting unit 402, a sound level superposition unit 403, and a dynamic error calibration unit 404.

[0170] Preferably, the signal processing unit 401 is used to acquire the original sound signal, perform analog-to-digital conversion on the original sound signal, and preprocess the converted sound signal to obtain the signal spectrum.

[0171] Preferably, the signal processing unit 401 preprocesses the analog-to-digital converted acoustic signal to obtain the signal spectrum, including:

[0172] Anti-aliasing filtering is applied to the acoustic signal after analog-to-digital conversion.

[0173] The signal after anti-aliasing filtering is framed, a short-time Fourier transform is performed on each frame, and the proportion of low-frequency components within the frame is calculated.

[0174] The window function is determined based on the proportion of low-frequency components within the frame.

[0175] The window function is applied to the signal frame, and a short-time Fourier transform is performed to obtain the signal spectrum.

[0176] Preferably, the signal processing unit 401 calculates the proportion η of intra-frame low-frequency components using the following method:

[0177] η = (Total sound energy in the 20Hz to 1kHz frequency band) / (Total sound energy in the entire 20Hz to 20kHz frequency band)

[0178] Wherein, using E(f)=|X(f)| 2 Calculate the sound energy, where f is the frequency of the sound signal and X(f) is the result of the short-time Fourier transform.

[0179] Preferably, the signal processing unit 401 determines a window function based on the proportion of low-frequency components within the frame, including:

[0180] If the proportion of low-frequency components within the frame is η≥30%, then the Blackman window is selected as the window function; if the proportion of low-frequency components within the frame is η<30%, then the Hanning window is selected as the window function.

[0181] Preferably, the segmented weighting unit 402 is used to segment the signal spectrum and perform weighting on each segment to obtain a first weighted sound level and a second weighted sound level.

[0182] Preferably, the segmented weighting unit 402 segments the signal spectrum and weights each segment to obtain a first weighted sound level and a second weighted sound level, including:

[0183] The signal spectrum is divided into two segments based on a dividing frequency of 1kHz.

[0184] For low-frequency signal spectra below 1kHz, a modified fast Fourier transform method is used for weighting to obtain the first weighted sound level; for mid-to-high frequency signal spectra of 1kHz and above, an optimized infinite impulse response digital filter method is used for weighting to obtain the second weighted sound level.

[0185] Preferably, the segmented weighting unit 402 employs a modified fast Fourier transform method for weighting to obtain a first weighted sound level, including:

[0186] Zero-padding is applied to the spectrum of low-frequency signals.

[0187] For any frequency point f located between adjacent spectral lines, use its left and right adjacent spectral lines f k f k+1 Weighted sound level L k L k+1 and its corresponding sound energy E k E k+1 The acoustic energy E(f) at point f is calculated using cubic spline interpolation.

[0188] The acoustic energy E(f) at each frequency point f is converted into the corresponding weighted sound level to obtain the first weighted sound level.

[0189] Preferably, the segmented weighting unit 402 employs an optimized infinite impulse response digital filter method for weighting to obtain a second weighted sound level, including:

[0190] An infinite impulse response digital filter quantized with 32-bit floating-point coefficients is used to perform real-time recursive filtering on the spectrum of mid-to-high frequency signals to obtain a weighted time-domain signal.

[0191] The short-time power spectrum of the time-domain signal is calculated and converted into acoustic energy.

[0192] The sound energy at each frequency point is converted into the corresponding weighted sound level to obtain the second weighted sound level;

[0193] Specifically, the pre-distortion formula in the bilinear transform is modified by introducing a pole fine-tuning coefficient k, thus revising the pre-distortion formula to Ω = 2kf. s tan(πf / f s This is to compensate for the amplitude-frequency response deviation caused by the bilinear transform; where Ω is the filter angular frequency; k is the pole fine-tuning coefficient; fs is the acoustic signal sampling frequency; and f is the acoustic signal frequency.

[0194] An infinite impulse response digital filter is implemented by cascading multiple second-order direct type II networks. The A-weighted filter uses three cascaded second-order direct type II networks, with the transfer function being... The C-weighted filter uses two cascaded second-order direct type II networks, with the transfer function being... Perform 32-bit floating-point quantization on the filter coefficients; where c and d represent the relevant undetermined coefficients, and z is the transform sign.

[0195] Preferably, the sound level superposition unit 403 is used to superimpose the first weighted sound level and the second weighted sound level to determine the total weighted sound level.

[0196] Preferably, the sound level superposition unit 403 superimposes the first weighted sound level and the second weighted sound level to determine the total weighted sound level, including:

[0197] L total =10lg(E low +E high ),

[0198]

[0199] Among them, L total For total weighted sound level; L low (f) represents the first weighted sound level; L high (f) represents the second weighted sound level; E low E represents the total acoustic energy in the low-frequency range. high This represents the total acoustic energy in the high-frequency band.

[0200] Preferably, the dynamic error calibration unit 404 is used to perform dynamic error calibration on the total weighted sound level, determine the standard error, and output the total weighted sound level when the standard error is less than or equal to a preset error threshold.

[0201] Preferably, the dynamic calibration unit 404 performs dynamic error calibration on the total weighted sound level to determine the standard error, including:

[0202] Select key frequency points, output sinusoidal signals with a peak value of 1V and a frequency corresponding to each selected key frequency point through a standard signal generator, and calculate the weighted output value L. meas (f) Based on the weighted output value and the standard weighted value L std (f) Calculate the calibration coefficient δ(f) = L std (f)-L meas (f), and establish a table showing the correspondence between the calibration coefficient δ and the frequency f;

[0203] For each frequency point f in the signal to be weighted, the corresponding calibration coefficient δ(f) is obtained from the calibration coefficient-frequency correspondence table through linear interpolation, and the uncalibrated weighted sound level L is then... raw (f) Make corrections to obtain the calibrated sound level L. cal (f)=L raw (f)+δ(f);

[0204] A standard signal with a frequency of 1kHz and a peak-to-peak value of 1V is generated using a standard signal generator. Single-point frequency verification is performed to determine the calibration error ΔL = L. raw (f)+|L cal (1kHz)-0|.

[0205] Preferably, the dynamic calibration unit 404 is further configured to:

[0206] If the calibration error is greater than the preset error threshold, the calibration coefficients of all key frequency points are remeasured and the correspondence table between the calibration coefficient δ and the frequency f is updated.

[0207] The hybrid high-precision frequency weighting system 400 of this invention corresponds to the hybrid high-precision frequency weighting method 100 of another embodiment of this invention, and will not be described again here.

[0208] According to another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of a hybrid high-precision frequency weighting method.

[0209] According to another aspect of the present invention, the present invention provides an electronic device, comprising:

[0210] The aforementioned computer-readable storage medium; and

[0211] One or more processors for executing a program in the computer-readable storage medium.

[0212] The present invention has been described with reference to a few embodiments. However, it will be apparent to those skilled in the art that other embodiments besides those disclosed above fall equivalently within the scope of the present invention.

[0213] Generally, all terms used in this invention are interpreted according to their ordinary meaning in the art, unless otherwise expressly defined herein. All references to “a / the / the [device, component, etc.]” ​​are openly interpreted as at least one instance of said device, component, etc., unless otherwise expressly stated. The steps of any method disclosed herein need not be performed in the exact order disclosed unless explicitly stated otherwise.

[0214] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0215] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0216] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0217] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0218] 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 it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A hybrid high-precision frequency weighting method, characterized in that, The method includes: The original acoustic signal is acquired, analog-to-digital conversion is performed on the original acoustic signal, and the converted acoustic signal is preprocessed to obtain the signal spectrum; The signal spectrum is segmented, and each segment is weighted to obtain a first weighted sound level and a second weighted sound level; The first weighted sound level and the second weighted sound level are superimposed to determine the total weighted sound level; The total weighted sound level is dynamically calibrated to determine the standard error, and the total weighted sound level is output when the standard error is less than or equal to a preset error threshold.

2. The method according to claim 1, characterized in that, The analog-to-digital converted audio signal is preprocessed to obtain the signal spectrum, including: Anti-aliasing filtering is applied to the acoustic signal after analog-to-digital conversion. The signal after anti-aliasing filtering is framed, a short-time Fourier transform is performed on each frame, and the proportion of low-frequency components within the frame is calculated. The window function is determined based on the proportion of low-frequency components within the frame. The window function is applied to the signal frame, and a short-time Fourier transform is performed to obtain the signal spectrum.

3. The method according to claim 2, characterized in that, The method calculates the proportion η of intra-frame low-frequency components using the following methods: η = (Total sound energy in the 20Hz to 1kHz frequency band) / (Total sound energy in the entire 20Hz to 20kHz frequency band) Where, using E(f)=|X(f)| 2 Calculate the sound energy, where f is the frequency of the sound signal and X(f) is the result of the short-time Fourier transform.

4. The method according to claim 2, characterized in that, The window function is determined based on the proportion of low-frequency components within the frame, including: If the proportion of low-frequency components within the frame is η≥30%, then the Blackman window is selected as the window function; if the proportion of low-frequency components within the frame is η<30%, then the Hanning window is selected as the window function.

5. The method according to claim 1, characterized in that, The signal spectrum is segmented, and each segment is weighted to obtain a first weighted sound level and a second weighted sound level, including: The signal spectrum is divided into two segments based on a dividing frequency of 1kHz. For low-frequency signal spectra below 1kHz, a modified fast Fourier transform method is used for weighting to obtain the first weighted sound level; for mid-to-high frequency signal spectra of 1kHz and above, an optimized infinite impulse response digital filter method is used for weighting to obtain the second weighted sound level.

6. The method according to claim 5, characterized in that, A modified Fast Fourier Transform method is used for weighting to obtain the first weighted sound level, including: Zero-padding is applied to the spectrum of low-frequency signals. For any frequency point f located between adjacent spectral lines, use its left and right adjacent spectral lines f k f k+1 Weighted sound level L k L k+1 and its corresponding sound energy E k E k+1 The acoustic energy E(f) at point f is calculated using cubic spline interpolation. The acoustic energy E(f) at each frequency point f is converted into the corresponding weighted sound level to obtain the first weighted sound level.

7. The method according to claim 5, characterized in that, An optimized infinite impulse response digital filter method is used for weighting to obtain the second weighted sound level, including: An infinite impulse response digital filter quantized with 32-bit floating-point coefficients is used to perform real-time recursive filtering on the spectrum of mid-to-high frequency signals to obtain a weighted time-domain signal. The short-time power spectrum of the time-domain signal is calculated and converted into acoustic energy. The sound energy at each frequency point is converted into the corresponding weighted sound level to obtain the second weighted sound level; Specifically, the pre-distortion formula in the bilinear transform is modified by introducing a pole fine-tuning coefficient k, thus revising the pre-distortion formula to Ω = 2kf. s tan(πf / f s This is to compensate for the amplitude-frequency response deviation caused by the bilinear transform; where Ω is the filter angular frequency; k is the pole fine-tuning coefficient; fs is the acoustic signal sampling frequency; and f is the acoustic signal frequency. An infinite impulse response digital filter is implemented by cascading multiple second-order direct type II networks. The A-weighted filter uses three cascaded second-order direct type II networks, with the transfer function being... The C-weighted filter uses two cascaded second-order direct type II networks, with the transfer function being... Perform 32-bit floating-point quantization on the filter coefficients; where c and d represent the relevant undetermined coefficients, and z is the transform sign.

8. The method according to claim 1, characterized in that, The first weighted sound level and the second weighted sound level are superimposed to determine the total weighted sound level, including: L total =10lg(E low +E high ), Among them, L total For total weighted sound level; L low (f) represents the first weighted sound level; L high (f) represents the second weighted sound level; E low E represents the total acoustic energy in the low-frequency range. high This represents the total acoustic energy in the high-frequency band.

9. The method according to claim 1, characterized in that, The total weighted sound level is subjected to dynamic error calibration to determine the standard error, including: Select key frequency points, output sinusoidal signals with a peak value of 1V and a frequency corresponding to each selected key frequency point through a standard signal generator, and calculate the weighted output value L. meas (f) Based on the weighted output value and the standard weighted value L std (f) Calculate the calibration coefficient δ(f) = L std (f)-L meas (f), and establish a table showing the correspondence between the calibration coefficient δ and the frequency f; For each frequency point f in the signal to be weighted, the corresponding calibration coefficient δ(f) is obtained from the calibration coefficient-frequency correspondence table through linear interpolation, and the uncalibrated weighted sound level L is then... raw (f) Make corrections to obtain the calibrated sound level L. cal (f)=L raw (f)+δ(f); A standard signal with a frequency of 1kHz and a peak-to-peak value of 1V is generated using a standard signal generator. Single-point frequency verification is performed to determine the calibration error ΔL = L. raw (f)+|L cal (1kHz)-0|.

10. The method according to claim 9, characterized in that, The method further includes: If the calibration error is greater than the preset error threshold, the calibration coefficients of all key frequency points are remeasured and the correspondence table between the calibration coefficient δ and the frequency f is updated.

11. A hybrid high-precision frequency weighting system, characterized in that, The system includes: The signal processing unit is used to acquire the original acoustic signal, perform analog-to-digital conversion on the original acoustic signal, and preprocess the analog-to-digital converted acoustic signal to obtain the signal spectrum; The segmented weighting unit is used to segment the signal spectrum and perform weighting on each segment to obtain a first weighted sound level and a second weighted sound level. A sound level superposition unit is used to superimpose the first weighted sound level and the second weighted sound level to determine the total weighted sound level; The dynamic error calibration unit is used to perform dynamic error calibration on the total weighted sound level, determine the standard error, and output the total weighted sound level when the standard error is less than or equal to a preset error threshold.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-10.

13. An electronic device, characterized in that, include: The computer-readable storage medium as described in claim 13; as well as One or more processors for executing a program in the computer-readable storage medium.

Images