Visualization of three-dimensional audio data as frequency-angle 3D model and data processing method thereof

Abstract

The present invention relates to the field of audio and video signal processing and visualization technology, and specifically to a method for visualizing three-dimensional audio data as a frequency-domain-angle 3D model and a data processing method thereof: Compared with traditional index testing, the present invention combines data processing and three-dimensional visualization schemes, and through modeling the three-dimensional stereo field of a speaker system, smoothing and filtering the three-dimensional frequency response curve, and establishing a directional map of the speaker system, it is possible to more intuitively evaluate the performance of the speaker in space, especially its frequency response curve and directivity, and to simply and clearly present the size relationship of the frequency response curve of the speaker system in various directions. At the same time, the present invention can be combined with traditional index testing schemes to evaluate the performance of the speaker system from different angles.

Description

Technical Field

[0001] The present invention relates to the technical field of audio and video signal processing and visualization, and in particular to a method for visualizing three-dimensional audio data into a frequency domain-angle 3D model and a data processing method thereof. Background Art

[0002] In modern society, with the advancement of technology and the improvement of quality of life, the public's requirements for sound quality are getting higher and higher. Whether it is a home theater system, virtual reality equipment, mobile communication tools, car audio or professional audio equipment, high-quality sound effects are essential. In these systems, speakers play a vital role, and the quality of their performance directly determines the overall sound quality of the audio and video system. Therefore, accurate performance evaluation of speakers is crucial for consumers to choose suitable products, and it also promotes the development of audio technology and the improvement of quality. In addition, accurate speaker evaluation can help manufacturers gain an in-depth understanding of the specific impact of different designs and material choices on sound quality, so as to make more effective adjustments and optimizations during the product development stage.

[0003] However, traditional speaker sound quality evaluation methods, such as hum detection, total harmonic distortion plus noise, pitch ratio, prominence ratio, and signal-to-noise ratio, are mainly performed by measuring specific single indicators of audio signals. Most of these methods focus on evaluating the response of a single sound source point and cannot provide a comprehensive and intuitive view of the distribution of the entire spatial sound field. Summary of the invention

[0004] The purpose of the present invention is to provide a method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and its data processing method, so as to solve the problem that traditional speaker sound quality evaluation methods mostly focus on evaluating the response of a single sound source point and cannot provide a comprehensive and intuitive view of the entire spatial sound field distribution.

[0005] To achieve the above object, the present invention provides a method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof, wherein the steps of visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof are as follows:

[0006] S0, select the product to be tested and measure its maximum linear unit;

[0007] S1. Determine the far field of the loudspeaker according to the far field definition criteria;

[0008] S2, placing the microphone test group within the far field range and arranging them according to the angular resolution;

[0009] S3. In a free field environment, a pulse signal is played with a loudspeaker and the pulse signal is picked up with a microphone;

[0010] S4, setting the sampling rate according to Shannon's theorem, and resampling the picked-up audio signal according to the sampling rate;

[0011] S5. To prevent spectrum leakage, select a time window function according to actual test requirements;

[0012] S6, converting the time domain signal into a frequency domain signal by Fourier transform;

[0013] S7, dynamically expand frequency domain data according to free field conditions and required measurement radius;

[0014] S8, setting the amplitude of the frequency band less than 20 Hz and greater than 20 kHz in the frequency domain to 0;

[0015] S9, performing octave filtering based on octave parameters, frequency array and time-domain sound pressure level array to smooth the original curve;

[0016] S10, converting the sound pressure in the frequency domain into a sound pressure level according to the reference sound pressure;

[0017] S11, generating an angle array according to the angle resolution during measurement;

[0018] S12, normalizing the sound pressure level data obtained after smoothing;

[0019] S13, according to the relationship between the data sets, correspond the angle array, the frequency array and the sound pressure level array, and draw a three-dimensional spectrum line diagram;

[0020] S14, using the drawing system to draw a three-dimensional spectrum rendering in the horizontal direction and the vertical direction, and smoothing the color value and filling the gradient color;

[0021] S15. Draw the top view of the spectrum rendering separately to obtain the directivity diagram of the speaker.

[0022] In step S5: the window function of the Hanning window is: Where ω(n) is the value of the window function at the nth point, N is the length of the window (i.e. the total number of points in the window), and n is the sequence number of the current point, from 0 to N-1.

[0023] In step S6: the function of continuous Fourier transform is: Where f(t) represents the continuous input signal, F(w) represents the continuous Fourier transform, ω is the continuous frequency, k is the discrete frequency index, and j is the imaginary unit.

[0024] In step S7: the formula for free field radius adjustment is: Among them, P(f,r2,φ,θ) represents the complex sound pressure at the distance r2, azimuth θ and elevation Φ. Similarly, P(f,r1,φ,θ) represents the complex sound pressure at the distance r1, azimuth θ and elevation Φ, where the wave number k = 2Πf / c.

[0025] In step S9: the smoothing algorithm for the octave is a smoothing algorithm based on one-third octave octave. The smoothing algorithm uses the frequency in the frequency array as the center frequency and calculates the upper and lower limits of the frequency respectively, wherein the lower limit of the frequency is: The frequency cap is: And the frequency domain data between the upper and lower limits in the frequency array are averaged to represent the actual sound pressure level of the center frequency, that is, Where sum is the sum of the sound pressure levels between the upper and lower frequency limits, and count is the number of points between the upper and lower frequency limits.

[0026] In step S10: the sound pressure level is calculated as: Among them, L p represents the sound pressure level, p represents the sound pressure, p ref Represents the reference sound pressure, which is 20 microPa.

[0027] In step S12: the sound pressure level in other directions needs to be normalized first: dBSPL = dBSPL-dBSPL ref Then normalize the sound pressure level in the reference direction to 0, i.e. dBSPL ref =0.

[0028] The present invention visualizes three-dimensional audio data as a frequency domain-angle 3D model and its data processing method. Compared with the traditional index test, the present invention combines data processing and three-dimensional visualization schemes. By modeling the three-dimensional stereo field of the speaker system, smoothing and filtering the three-dimensional frequency response curve, and establishing the directional map of the speaker system, the performance of the speaker in space, especially its frequency response curve and directivity, can be more intuitively evaluated, and the size relationship of the frequency response curve of the speaker system in various directions can be simply and clearly presented. At the same time, the present invention can be combined with the traditional index test scheme to evaluate the performance of the speaker system from different angles. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0030] Figure 1It is a specific flow chart of the method for visualizing three-dimensional audio data into a frequency domain-angle 3D model and data processing thereof provided by the present invention.

[0031] Figure 2 It is a pulse signal diagram in a specific embodiment of the present invention.

[0032] Figure 3 In the specific embodiment of the present invention, Figure 2 The three-dimensional spectrum obtained by playing the pulse signal.

[0033] Figure 4 The present invention provides Figure 3 One-third octave smoothed spectrum of the three-dimensional spectrum.

[0034] Figure 5 In the specific embodiment of the present invention, Figure 2 The horizontal spectrum rendering of the played pulse signal.

[0035] Figure 6 In the specific embodiment of the present invention, Figure 2 Vertical spectrum rendering of the played pulse signal.

[0036] Figure 7 In the specific embodiment of the present invention, Figure 2 Directivity map of the pulse signal played. DETAILED DESCRIPTION

[0037] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0038] See also Figures 1 to 7 The present invention provides a method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and a data processing method thereof. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and a data processing method thereof comprises the following steps:.

[0039] S0, measure the maximum linear unit of the single speaker (i.e. the longest side of the speaker), and record it as L;

[0040] S1. Calculate the far field critical radius R according to the far field definition rules far , and select the radius of data collection r = 1m according to this radius;

[0041] S2. Place the microphone in the far field, here select r = 1m, and set the angle resolution to 5 degrees;

[0042] S3. In the free field environment guaranteed by the fully anechoic room, use the speaker to be tested to play the pulse signal, pick up the pulse signal through the microphone, and name the pulse signals in each direction according to certain rules;

[0043] S4. Set the sampling rate to fs = 48000 Hz, and resample the time domain signals of each direction just picked up according to this sampling rate. The resampled data should include the time array Times, the real part array Real of the sound pressure, and the imaginary part array Imag of the sound pressure;

[0044] S5. Add a Hanning window with an effective length of 7.6ms to all the above data files. The length of the window function should be enough to limit the time of the first reflection sound. In the part outside the window function, the frequency range of 20-20k should be filled with 0;

[0045] S6, performing Fourier transform on the above-mentioned resampled and windowed time domain files to obtain frequency domain data at various angles. After this step, each frequency domain file should include the following three parts: frequency array Freqs, sound pressure amplitude array Magn, and phase array Phase;

[0046] S7. Since the radius to be measured selected here is consistent with the measured radius, the data is no longer expanded and the above data is kept unchanged;

[0047] S8. According to specific implementation requirements, the cutoff frequency is set to 100 Hz, and all data points outside 100 Hz-20000 Hz are normalized to 0;

[0048] S9, performing octave smoothing on the sound pressure amplitude array according to the octave parameter Octave and the frequency domain array Freqs;

[0049] S10, converting the interpolated array from sound pressure to sound pressure level, with a reference sound pressure of 20 μPa;

[0050] S11. Generate an azimuth array Angle according to the angular resolution of 5 degrees during data collection. The azimuth array should cover 180 degrees. Note that different spectrum renderings require different data.

[0051] S12. After the above steps, all data for three-dimensional drawing have been obtained. In order to measure the relationship between the sound pressure in each direction and the sound pressure on the axis, the data set is normalized by the sound pressure on the axis, that is, the sound pressure on the axis is subtracted from the data in all directions.

[0052] S13. According to the relationship between the data sets, the angle array, the frequency array and the sound pressure level array are matched, and a three-dimensional spectrum line graph is drawn. At the same time, the algorithm proposed in the back is used to optimize the display effect, and a color gradient value of a single angle curve is added to the spectrum line graph, so as to highlight the frequency response curve of a certain angle in the later stage;

[0053] S14, mapping color values ​​according to the sound pressure level, drawing horizontal spectrum graphs and vertical spectrum graphs, using GDI+ to smooth the color display, and filling the color gradient;

[0054] S15. Render the top view of the spectrum graph separately as a directivity graph of the loudspeaker, and add a dynamic adjustment function to it to normalize the display of the three-dimensional graph.

[0055] In this embodiment, compared with the traditional index test, the present invention combines data processing and three-dimensional visualization schemes, and through the three-dimensional stereo field modeling of the speaker system, the smoothing filtering of the three-dimensional frequency response curve, and the establishment of the directional map of the speaker system, the performance of the speaker in space, especially its frequency response curve and directivity, can be more intuitively evaluated, and the magnitude relationship of the frequency response curve of the speaker system in various directions can be simply and clearly presented. At the same time, the present invention can be combined with the traditional index test scheme to evaluate the performance of the speaker system from different angles.

[0056] The far field definition in step S1 is closely related to the longest line length L measured in step S0. far Need to meet condition 1: R far >>L; Condition 2: R far >>λ; Condition 3: Among them, condition one requires that the test distance is greater than the maximum geometric dimension, condition two requires that the distance is greater than the wavelength, and condition three is mainly applicable to large speaker systems (such as line arrays, sound bars and sound boards).

[0057] In step S3 , it is required to perform measurement in a free sound field, where the free sound field should use an anechoic room that meets corresponding standards.

[0058] The dynamic adjustment of frequency domain data in step S7 means that, for the convenience of measurement, the original signal collected by the microphone may not be based on the target radius. At this time, according to the propagation law of sound in the free field, the data can be expanded within the free field radius according to the inverse law of sound pressure with distance.

[0059] The smoothing algorithm for the octave in step S9 is based on a smoothing algorithm for one-third octave octaves.

[0060] In step S10, the reference sound pressure corresponds to the minimum sound pressure that the human ear can perceive at a frequency of 1000 Hz, which is 20 μPa (20×10-6 Pa).

[0061] In step S11, the angular resolution may refer to 1 degree, 2.5 degrees, 5 degrees, 10 degrees or other suitable resolutions, wherein the preferred resolution is a common factor of 90, 180, and 360. In actual testing, the angle value may be an integer, which is beneficial to the subsequent data processing based on a computer program. Furthermore, since the actual measurement space includes azimuth and elevation angles, one of which has a greater influence on the sound pressure level, when drawing, only the angle in that direction needs to be extracted as an angle array, and the other angle is generally locked to 0 degrees (horizontal rendering) or 90 degrees (vertical rendering).

[0062] In step S12, normalization refers to normalizing data at different angles at the same frequency to a certain value, and the axial sound pressure level or the maximum sound pressure level is selected. The so-called axial direction refers to the direction facing the speaker diaphragm. In most cases, this is not much different from normalizing to the maximum value, because the axial direction is generally the direction in which sound propagates the loudest.

[0063] In step S13, in order to dynamically observe the concave and convex relationship of the spectrum graph, it is necessary to dynamically adjust the coordinate axis range according to the needs. At the same time, the sound pressure level that is not in the coordinate axis range after adjustment is normalized to 0, which is conducive to analyzing the size relationship of the entire spectrum graph.

[0064] In step S14, the horizontal spectrum rendering diagram and the vertical spectrum rendering diagram are divided according to another axis other than the direction axis.

[0065] The formula provided by this method is as follows:

[0066] In step S5: the window function of the Hanning window is: Where ω(n) is the value of the window function at the nth point, N is the length of the window (i.e. the total number of points in the window), and n is the sequence number of the current point, from 0 to N-1.

[0067] In step S6: the function of continuous Fourier transform is: Where f(t) represents the continuous input signal, F(w) represents the continuous Fourier transform, ω is the continuous frequency, k is the discrete frequency index, and j is the imaginary unit.

[0068] In step S7: the formula for free field radius adjustment is: Among them, P(f,r2,φ,θ) represents the complex sound pressure at the distance r2, azimuth θ and elevation Φ. Similarly, P(f,r1,φ,θ) represents the complex sound pressure at the distance r1, azimuth θ and elevation Φ, where the wave number k = 2Πf / c.

[0069] In step S9: the smoothing algorithm uses the frequency in the frequency array as the center frequency and calculates the upper and lower limits of the frequency respectively, where the lower limit of the frequency is: The frequency cap is: And the frequency domain data between the upper and lower limits in the frequency array are averaged to represent the actual sound pressure level of the center frequency, that is, Where sum is the sum of the sound pressure levels between the upper and lower frequency limits, and count is the number of points between the upper and lower frequency limits.

[0070] In step S10: the sound pressure level is calculated as: Among them, L p represents the sound pressure level, p represents the sound pressure, p ref Represents the reference sound pressure, which is 20 microPa.

[0071] In step S12: the sound pressure level in other directions needs to be normalized first: dBSPL = dBSPL-dBSPL ref Then normalize the sound pressure level in the reference direction to 0, i.e. dBSPL ref =0.

[0072] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A method for visualizing three-dimensional audio data into a frequency domain-angle 3D model and data processing method thereof, characterized in that: Here are the steps: S0, select the product to be tested and measure its maximum linear unit; S1. Determine the far field of the loudspeaker according to the far field definition criteria; S2, placing the microphone test group within the far field range and arranging them according to the angular resolution; S3. In a free field environment, a pulse signal is played with a loudspeaker and the pulse signal is picked up with a microphone; S4, setting the sampling rate according to Shannon's theorem, and resampling the picked-up audio signal according to the sampling rate; S5. To prevent spectrum leakage, select a time window function according to actual test requirements; S6, converting the time domain signal into a frequency domain signal by Fourier transform; S7, dynamically expand frequency domain data according to free field conditions and required measurement radius; S8, setting the amplitude of the frequency band less than 20 Hz and greater than 20 kHz in the frequency domain to 0; S9, performing octave filtering based on octave parameters, frequency array and time-domain sound pressure level array to smooth the original curve; S10, converting the sound pressure in the frequency domain into a sound pressure level according to the reference sound pressure; S11, generating an angle array according to the angle resolution during measurement; S12, normalizing the sound pressure level data obtained after smoothing; S13, according to the relationship between the data sets, correspond the angle array, the frequency array and the sound pressure level array, and draw a three-dimensional spectrum line diagram; S14, using the drawing system to draw a three-dimensional spectrum rendering in the horizontal direction and the vertical direction, and smoothing the color value and filling the gradient color; S15. Draw the top view of the spectrum rendering separately to obtain the directivity diagram of the speaker.

2. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 1, characterized in that: In step S5: the window function of the Hanning window is: Where ω(n) is the value of the window function at the nth point, N is the length of the window, and n is the sequence number of the current point, from 0 to N-1.

3. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 2, characterized in that: In step S6: the function of continuous Fourier transform is: Where f(t) represents the continuous input signal, F(w) represents the continuous Fourier transform, ω is the continuous frequency, k is the discrete frequency index, and j is the imaginary unit.

4. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 3, characterized in that: In step S7: the formula for free field radius adjustment is: Among them, P(f,r2,φ,θ) represents the complex sound pressure at the distance r2, azimuth θ and elevation Φ. Similarly, P(f,r1,φ,θ) represents the complex sound pressure at the distance r1, azimuth θ and elevation Φ, where the wave number k = 2Πf / c.

5. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 4, characterized in that: In step S9: the smoothing algorithm for the octave is a smoothing algorithm based on one-third octave octave. The smoothing algorithm uses the frequency in the frequency array as the center frequency and calculates the upper and lower limits of the frequency respectively, wherein the lower limit of the frequency is: The frequency cap is:

6. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 5, characterized in that: In step S9: the frequency domain data between the upper and lower limits in the frequency array are averaged to represent the actual sound pressure level of the center frequency, that is, Where sum is the sum of the sound pressure levels between the upper and lower frequency limits, and count is the number of points between the upper and lower frequency limits.

7. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 6, characterized in that: In step S10: the sound pressure level is calculated as: Among them, L p represents the sound pressure level, p represents the sound pressure, p ref Represents the reference sound pressure, which is 20 microPa.

8. The method for visualizing three-dimensional audio data as a frequency domain-angle 3D model and data processing thereof as claimed in claim 7, characterized in that: In step S12: the sound pressure level in other directions needs to be normalized first: dBSPL = dBSPL-dBSPL ref Then normalize the sound pressure level in the reference direction to 0, i.e. dBSPL ref =0.

Images