Sound box sound effect intelligent adjusting method based on data analysis

Abstract

The invention discloses a sound box sound effect intelligent adjustment method based on data analysis, and relates to the field of sound box sound effect adjustment. The method comprises the following steps: classifying use modes of a to-be-adjusted sound box according to different use scenes, extracting acoustic characteristic parameters of a test audio and a sampling audio in different use modes, calculating the difference degree of the j acoustic characteristic parameter of the sampling audio and the test audio in the i use mode, obtaining influence factor data of the difference degree in the i use mode, constructing a neural network model taking the influence factor data as an input parameter and taking the difference degree as an output parameter, collecting real-time data of the kth influence factor at the current moment when the sound equipment to be adjusted works in the ith use mode, and obtaining the difference degree of acoustic characteristic parameters corresponding to the real-time data by utilizing a trained neural network, and calculating a correction coefficient of the jth acoustic characteristic parameter xij of the real-time audio played at the current moment, and dynamically adjusting the acoustic characteristic parameter of the real-time audio at the next moment by using the correction coefficient.

Description

Technical Field

[0001] The present invention relates to the field of sound and sound effect adjustment, and in particular to a sound and sound effect intelligent adjustment method based on data analysis. Background Art

[0002] With the development of automotive electronics, car audio and entertainment systems are becoming more and more intelligent and networked. Since the expressiveness of the sound system directly affects the driving experience of the smart cockpit, car consumers are gradually paying more and more attention to the smart cockpit sound system.

[0003] The sound effects of the car audio system are not only directly affected by the design of the speaker itself, the internal structure, material, size and other factors of the speaker, but also by the sound of the use environment. When the vehicle is driving, as the driving state changes, the driving environment is also constantly changing. Under the combined influence of multiple external factors, there will be a large deviation between the sound emitted by the speaker and the sound heard by the people in the car. However, at present, the existing car audio system has not introduced a practical and effective sound effect adjustment method for this deviation, which affects the experience of users in the car. To this end, we propose an intelligent sound effect adjustment method based on data analysis. Summary of the invention

[0004] The main purpose of the present invention is to provide a method for intelligently adjusting sound effects based on data analysis, which can effectively solve the problems in the background technology.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for intelligently adjusting sound effects based on data analysis, comprising:

[0007] Classifying the usage modes of the audio to be adjusted into a first usage mode, a second usage mode, an nth usage mode, etc. according to different usage scenarios, allowing the audio to be adjusted to play test audio in different usage modes respectively, and sampling the test audio during the playing process to obtain sampled audio;

[0008] Extract the acoustic characteristic parameters of the test audio and the sampled audio respectively, and calculate the difference Δx between the jth acoustic characteristic parameter of the sampled audio and the test audio in the i-th usage mode ij The acoustic characteristic parameters include at least one of frequency, loudness, timbre, duration, amplitude, power sound pressure level, spectrum, echo and reverberation; the difference Δx ij The calculation formula is:

[0009]

[0010] In the formula, x ijis the jth acoustic characteristic parameter of the test audio in the i-th usage mode; x ij ' is the jth acoustic characteristic parameter of the sampled audio in the i-th usage mode; i = 1, 2, ..., n;

[0011] The factors affecting the difference in the acoustic characteristic parameters are analyzed to obtain the difference Δx in the i-th usage mode. ij Influencing factors data if ik , if ik For the kth influencing factor data under the i-th usage mode, a neural network model is constructed with the normalized influencing factor data as an input parameter and the normalized difference as an output parameter, and the neural network is trained using the acquired data until its prediction accuracy reaches an expected value;

[0012] The weight coefficient of the kth influencing factor data in the i-th usage mode in the input parameters of the neural network is defined as , weight coefficient The determination process includes the following steps:

[0013] For the difference Δx ij If the influencing factor data ik Perform grey correlation analysis to obtain the kth influencing factor data and the difference Δx under the i-th usage mode ij The correlation between

[0014] The degree of association Normalization is performed to obtain the normalized correlation As the weight coefficient of the kth influencing factor data in the i-th usage mode in, Where K is the number of influencing factors in the i-th usage mode;

[0015] The calculation formula of the expected value is:

[0016]

[0017] Among them, E(Y) represents the expected value; N represents the number of neural network input samples; f(X i ) is represented as the output function of the neural network; X i Represented as the i-th output sample of the neural network;

[0018] The data normalization formula is:

[0019]

[0020] In the formula, X' is the data value after normalization; X is the value before normalization; X min is the minimum value of the data before normalization; X max is the maximum value of the data before normalization;

[0021] Collect the real-time data of the kth influencing factor at the current moment when the sound to be adjusted is working in the ith usage mode Using trained neural networks to obtain real-time data if ik The difference of the corresponding j-th acoustic characteristic parameter

[0022] According to the difference obtained Calculate the jth acoustic characteristic parameter of the real-time audio played at the current moment when the sound to be adjusted is working in the i-th usage mode Correction factor k ij , using the correction factor k ij Dynamically adjust the acoustic characteristic parameters of the real-time audio at the next moment to obtain the corrected acoustic characteristic parameters of the real-time audio , and output the corrected real-time audio, the correction coefficient k ij The calculation formula is: Modified acoustic characteristic parameters for real-time audio The calculation formula is:

[0023] The present invention has the following beneficial effects:

[0024] Compared with the prior art, the technical solution of the present invention classifies the usage modes of the sound to be adjusted according to different usage scenarios, and makes the sound to be adjusted play the test audio in different usage modes respectively, and samples the test audio during the playback process to obtain the sampled audio, respectively extract the acoustic feature parameters of the test audio and the sampled audio, and calculate the difference Δx between the j-th acoustic feature parameter of the sampled audio and the test audio in the i-th usage mode. ij , analyzing the influencing factors that cause the acoustic characteristic parameters to differ, and obtaining the difference Δx in the i-th usage mode. ij Influencing factors data if ik , if ik The kth influencing factor data in the i-th usage mode is constructed, and a neural network model is constructed with the normalized influencing factor data as input parameters and the normalized difference as output parameters. The neural network is trained with the acquired data until its prediction accuracy reaches the expected value. When the sound to be adjusted works in the i-th usage mode, the real-time data of the kth influencing factor at the current moment is collected. ik, use the trained neural network to obtain real-time data if ik The difference Δx of the corresponding j-th acoustic characteristic parameter ij , according to the difference Δx obtained ij Calculate the jth acoustic characteristic parameter x of the real-time audio played at the current moment when the sound to be adjusted is working in the i-th usage mode ij Correction factor k ij , using the correction factor k ij Dynamically adjust the acoustic characteristic parameters of the real-time audio at the next moment to obtain the corrected acoustic characteristic parameters x of the real-time audio ij The corrected real-time audio is output, which can correct the deviation of the car audio system under different usage modes, thereby improving the user's experience of the car audio system. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The present invention is a flow chart of a method for intelligently adjusting sound effects based on data analysis. DETAILED DESCRIPTION

[0026] The present invention will be further described below in conjunction with specific implementation methods, wherein the accompanying drawings are only used for exemplary descriptions and represent only schematic diagrams rather than actual drawings, and should not be understood as limiting the present invention. In order to better illustrate the specific implementation methods of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product.

[0027] Example 1

[0028] The specific implementation process of the technical solution of the present invention includes the following steps:

[0029] Step 1: Classify the usage mode of the audio to be adjusted into a first usage mode, a second usage mode, and an nth usage mode according to different usage scenarios;

[0030] The usage scenarios can be classified into stationary state and driving state according to whether the vehicle is in driving state. The usage mode of the audio to be adjusted can be classified into a first usage mode in stationary state and a second usage mode in driving state. Specifically, the second usage mode can be further subdivided according to the vehicle driving speed. The classification principle is similar and no secondary subdivision is performed here.

[0031] In this embodiment, the technical solution of the present invention is described by taking the usage mode of the audio to be adjusted as the first usage mode in a static state as an example.

[0032] Step 2: Instruct the speaker to be adjusted to play the test audio in different usage modes respectively, and sample the test audio during the playing process to obtain the sampled audio;

[0033] Step 3: Extract the acoustic feature parameters of the test audio and the sampled audio respectively, and calculate the difference Δx between the jth acoustic feature parameter of the sampled audio and the test audio in the i-th usage mode ij The acoustic characteristic parameters include at least one of frequency, loudness, timbre, duration, amplitude, power sound pressure level, spectrum, echo and reverberation; the difference Δx ij The calculation formula is:

[0034]

[0035] In the formula, x ij is the jth acoustic characteristic parameter of the test audio in the i-th usage mode; x ij ' is the jth acoustic characteristic parameter of the sampled audio in the i-th usage mode; i = 1, 2, ..., n;

[0036] Step 4: Analyze the factors that affect the acoustic characteristic parameters and obtain the difference Δx in the i-th usage mode. ij Influencing factors data if ik , if ik is the kth influencing factor data under the i-th usage mode;

[0037] It should be noted that when the vehicle is stationary, the following reasons may cause differences in acoustic characteristic parameters: 1) Sound source intensity: Sound source intensity is the core factor that determines loudness. The greater the sound source intensity, the louder the sound produced. 2) Distance: Sound will gradually attenuate during propagation. The closer to the sound source, the louder the sound; conversely, the farther the distance, the lower the loudness. 3) Medium density: The propagation effect of sound waves in different media is different. In solids, sound waves propagate faster and lose less energy, so the loudness is higher; while in gases, sound waves propagate slower, lose more energy, and have lower loudness. 4) Environmental noise. 5) Spatial characteristics: including hard surfaces, high ceilings, multiple noise sources, etc. Hard surfaces cause sound waves to bounce back and produce echoes; high ceilings prolong the propagation time of sound waves and produce echoes; multiple noise sources increase the complexity of the sound.

[0038] For data on influencing factors such as sound source intensity, distance, and environmental noise, their quantitative values ​​can be directly collected. For medium density, generally speaking, the medium in the vehicle interior is air, and parameter types with an influencing effect can be selected for replacement, such as using the humidity value in the air as a quantitative indicator; for spatial characteristics, the airtightness of the vehicle can be used as a characterization factor for quantitative value determination.

[0039] Step 5: Construct a neural network model with the normalized influencing factor data as input parameters and the normalized difference as output parameters, and use the acquired data to train the neural network until its prediction accuracy reaches the expected value; the calculation formula of the expected value is,

[0040]

[0041] Among them, E(Y) represents the expected value; N represents the number of neural network input samples; f(X i ) is represented as the output function of the neural network; X i Represented as the i-th output sample of the neural network;

[0042] In this embodiment, taking the constructed neural network as a BP neural network as an example, the number of neurons in the input layer of the neural network is equal to the type of input influencing factor data, the number of neurons in the output layer is 1, and the number of neurons in the intermediate hidden layer can be determined according to an empirical formula, which is: Among them, m and n are the number of neurons in the input layer and output layer of the neural network respectively, and a is a constant coefficient of 1-9.

[0043] Step 6: Define the weight coefficient of the kth influencing factor data in the i-th usage mode in the input parameters of the neural network as Weight coefficient The determination process includes the following steps:

[0044] Difference Δx ij and influencing factor data if ik Perform grey correlation analysis to obtain the kth influencing factor data and the difference Δx under the i-th usage mode ij The correlation between ;

[0045] Degree of association Normalization is performed to obtain the normalized correlation As the weight coefficient of the kth influencing factor data in the i-th usage mode ,in, Where K is the number of influencing factors in the i-th usage mode;

[0046] Step 7: Collect the real-time data of the kth influencing factor at the current moment when the sound to be adjusted is working in the ith usage mode , using trained neural networks to obtain real-time data The difference of the corresponding j-th acoustic characteristic parameter ;

[0047] Step 8: Based on the difference obtained Calculate the jth acoustic characteristic parameter of the real-time audio played at the current moment when the sound to be adjusted is working in the i-th usage mode Correction factor k ij , using the correction factor k ij Dynamically adjust the acoustic characteristic parameters of the real-time audio at the next moment to obtain the corrected acoustic characteristic parameters of the real-time audio , and output the corrected real-time audio, the correction coefficient k ij The calculation formula is: Modified acoustic characteristic parameters for real-time audio The calculation formula is:

[0048] Example 2

[0049] In this embodiment, the technical solution of the present invention is described by taking the usage mode of the audio to be adjusted as the second usage mode in the driving state as an example.

[0050] The specific implementation process of the technical solution of the present invention includes the following steps:

[0051] Step 1: Instruct the speaker to be adjusted to play the test audio in different usage modes respectively, and sample the test audio during the playing process to obtain the sampled audio;

[0052] Step 2: Extract the acoustic feature parameters of the test audio and the sampled audio respectively, and calculate the difference Δx between the jth acoustic feature parameter of the sampled audio and the test audio in the i-th usage mode ij The acoustic characteristic parameters include at least one of frequency, loudness, timbre, duration, amplitude, power sound pressure level, spectrum, echo and reverberation; the difference Δx ij The calculation formula is:

[0053]

[0054] In the formula, x ij is the jth acoustic characteristic parameter of the test audio in the i-th usage mode; x ij ' is the jth acoustic characteristic parameter of the sampled audio in the i-th usage mode; i = 1, 2, ..., n;

[0055] Step 3: Analyze the factors that affect the acoustic characteristic parameters and obtain the difference Δx in the i-th usage mode. ij Influencing factors data if ik , if ik is the kth influencing factor data under the i-th usage mode;

[0056] It should be noted that the reasons for the difference in acoustic characteristic parameters when the vehicle is in driving state are in addition to those mentioned in Example 1: 1) Sound source intensity: Sound source intensity is the core factor that determines loudness. The greater the sound source intensity, the louder the sound produced. 2) Distance: Sound will gradually attenuate during propagation. The closer to the sound source, the louder the sound; conversely, the farther the distance, the lower the loudness. 3) Medium density: The propagation effect of sound waves in different media is different. In solids, sound waves propagate faster and lose less energy, so the loudness is higher; while in gases, sound waves propagate slower, lose more energy, and have lower loudness. 4) Environmental noise. 5) Spatial characteristics: including hard surfaces, high ceilings, multiple noise sources, etc. Hard surfaces will cause sound waves to bounce back and produce echoes; high ceilings will prolong the propagation time of sound waves and produce echoes; multiple noise sources will increase the complexity of sound. Also includes: 6) Driving speed. The higher the speed, the greater the environmental noise generated. 7) Vibration factors. Mainly including vibration amplitude, vibration frequency, etc.

[0057] Step 4: Construct a neural network model with the normalized influencing factor data as input parameters and the normalized difference as output parameters, and use the acquired data to train the neural network until its prediction accuracy reaches the expected value; the calculation formula of the expected value is,

[0058]

[0059] Among them, E(Y) represents the expected value; N represents the number of neural network input samples; f(X i ) is represented as the output function of the neural network; X i Represented as the i-th output sample of the neural network;

[0060] Step 5: Define the weight coefficient of the kth influencing factor data in the i-th usage mode in the input parameters of the neural network as , weight coefficient The determination process includes the following steps:

[0061] Difference Δx ij If the influencing factor data ik Perform grey correlation analysis to obtain the kth influencing factor data and the difference Δx under the i-th usage mode ij The correlation between ;

[0062] Degree of association Normalization is performed to obtain the normalized correlation As the weight coefficient of the kth influencing factor data in the i-th usage mode ,in, Where K is the number of influencing factors in the i-th usage mode;

[0063] Step 6: Collect the real-time data of the kth influencing factor at the current moment when the sound to be adjusted is working in the ith usage mode , using trained neural networks to obtain real-time data The difference of the corresponding j-th acoustic characteristic parameter

[0064] Step 7: Based on the difference obtained Calculate the jth acoustic characteristic parameter of the real-time audio played at the current moment when the sound to be adjusted is working in the i-th usage mode Correction factor k ij , using the correction factor k ij Dynamically adjust the acoustic characteristic parameters of the real-time audio at the next moment to obtain the corrected acoustic characteristic parameters of the real-time audio , and output the corrected real-time audio, the correction coefficient k ij The calculation formula is: Modified acoustic characteristic parameters for real-time audio The calculation formula is:

[0065] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for intelligently adjusting sound effects based on data analysis, characterized in that: include: Classifying the usage modes of the audio to be adjusted into a first usage mode, a second usage mode, an nth usage mode, etc. according to different usage scenarios, allowing the audio to be adjusted to play test audio in different usage modes respectively, and sampling the test audio during the playing process to obtain sampled audio; Extract the acoustic characteristic parameters of the test audio and the sampled audio respectively, and calculate the difference Δx between the jth acoustic characteristic parameter of the sampled audio and the test audio in the i-th usage mode ij ; The factors affecting the difference in the acoustic characteristic parameters are analyzed to obtain the difference Δx in the i-th usage mode. ij Influencing factors data if ik , if ik For the kth influencing factor data under the i-th usage mode, a neural network model is constructed with the normalized influencing factor data as an input parameter and the normalized difference as an output parameter, and the neural network is trained using the acquired data until its prediction accuracy reaches an expected value; Collect the real-time data of the kth influencing factor at the current moment when the sound to be adjusted is working in the ith usage mode Leverage trained neural networks to obtain real-time data The difference of the corresponding j-th acoustic characteristic parameter According to the difference obtained Calculate the jth acoustic characteristic parameter of the real-time audio played at the current moment when the sound to be adjusted is working in the i-th usage mode Correction factor k ij , using the correction factor k ij Dynamically adjust the acoustic characteristic parameters of the real-time audio at the next moment to obtain the corrected acoustic characteristic parameters of the real-time audio And output the corrected real-time audio.

2. The method for intelligently adjusting sound effects based on data analysis according to claim 1, characterized in that: Difference Δx ij The calculation formula is: In the formula, x ij The jth acoustic characteristic parameter of the test audio in the i-th usage mode; x ij ' is the jth acoustic characteristic parameter of the sampled audio in the i-th usage mode; i=1,2,...,n.

3. The method for intelligently adjusting sound effects based on data analysis according to claim 1, characterized in that: The weight coefficient of the kth influencing factor data in the i-th usage mode in the input parameters of the neural network is defined as Weight coefficient The determination process includes the following steps: For the difference Δx ij If the influencing factor data ik Perform grey correlation analysis to obtain the kth influencing factor data and the difference Δx under the i-th usage mode ij The correlation between The degree of association Normalization is performed to obtain the normalized correlation As the weight coefficient of the kth influencing factor data in the i-th usage mode in, Where K is the number of influencing factors in the i-th usage mode.

4. The method for intelligently adjusting sound effects based on data analysis according to claim 1, characterized in that: The acoustic characteristic parameters include at least one of frequency, loudness, timbre, duration, amplitude, power sound pressure level, spectrum, echo and reverberation.

5. The method for intelligently adjusting sound effects based on data analysis according to claim 1, characterized in that: The calculation formula of the expected value is: Among them, E(Y) represents the expected value; N represents the number of neural network input samples; f(X i ) is represented as the output function of the neural network; X i Represented as the i-th output sample of the neural network.

6. The method for intelligently adjusting sound effects based on data analysis according to claim 1, characterized in that: Correction factor k ij The calculation formula is: Modified acoustic characteristic parameters for real-time audio The calculation formula is:

Images