Method and system for masking noise

Abstract

The present invention discloses a method and system for masking noise. The method for masking noise includes the following steps: training an intelligent neural network model with a variety of "ideal noises" that can soothe people's body and mind; using the trained intelligent neural network model to identify the actual noise to be masked and form a parameter matching table, the parameter matching table includes noise types, time scaling parameters, and slope transformation parameters, and at the same time obtaining the amplitude ratio of the actual noise; according to the parameter matching table, combined with the user's personalized settings, obtaining the ideal noise type, time scaling parameter, and slope transformation parameter with the best masking effect; generating the best personalized noise masking sound according to the best-matched ideal noise type, time scaling parameter, slope transformation parameter, and the amplitude ratio of the actual noise.

Description

Technical Field

[0001] The present invention belongs to the technical field of audio processing, and particularly relates to a method and a system for masking noise.

Background Art

[0002] As one of the most common wearable electronic products in daily life, noise-canceling headphones have become an indispensable part of people's daily life. However, due to the limitations of their performance, no noise-canceling headphones can completely eliminate noise. Especially those continuous, harsh, and loud noises, such as the sound of a drill, the sound of pile driving, the sound of an alarm, the roar of a machine, the sound of snoring, etc. When people need to concentrate on learning, or hope to rest quietly, or do not want to be disturbed by such noises, this kind of sound is like a devil that lingers and makes people feel distracted. It is simply impossible to study or rest, and it may cause physical discomfort, and more seriously, it may cause hearing damage to people and even lead to various diseases of the nervous system and cardiovascular system.

[0003] How to provide a noise masking method and system to solve the above problems is what those skilled in the art need to solve.

Summary of the Invention

[0004] The purpose of the present invention is to provide a method and a system for masking noise, so as to solve the problem that existing noise-canceling headphones cannot completely eliminate some continuous, harsh, and loud noises due to the limitations of their performance.

[0005] To achieve the above purpose, the present invention provides a method for masking noise, including the following steps:

[0006] Training an intelligent neural network model with a variety of ideal noises. The trained intelligent neural network model identifies the input noise and forms a parameter matching table composed of multiple groups of data arranged in descending order of matching degree. Each group of data parameters in the parameter matching table includes noise type, time scaling parameter, and slope transformation parameter;

[0007] Using the trained intelligent neural network model to identify the actual noise, obtaining a parameter matching table of the actual noise arranged in descending order of matching degree. Each group of data parameters in the table includes noise type, time scaling parameter, and slope transformation parameter, and at the same time obtaining the amplitude ratio of the actual noise;

[0008] According to the parameter matching table, select a group of parameters that is consistent with the ideal noise category selected by the user and has the highest ranking among all types of this category in the parameter matching table as the parameters that best match the ideal noise. The parameters that best match include noise type, time scaling parameter, and slope transformation parameter;

[0009] The masking sound is generated by adjusting the ideal noise that matches according to the most matching noise type, time scaling parameter, slope transformation parameter, and the amplitude ratio of the actual noise.

[0010] According to the above main features, in the step of training the intelligent neural network model with a variety of ideal noises, it includes providing a total of k types of ideal noise sources of white noise, pink noise, and Brown noise. First, each ideal noise is normalized, and then each normalized ideal noise is respectively subjected to m types of time scaling processes to obtain k*m types of ideal noises. Then, each time-scaled ideal noise is subjected to n types of audio characteristic slope transformations to finally obtain k*m*n types of ideal noises; MFCC feature extraction is performed on each of the k*m*n types of ideal noises, and the MFCC feature parameters of the obtained k*m*n types of ideal noises are used as the training data of the neural network model to train the neural network model. The trained intelligent neural network model performs recognition calculation on the input noise and forms a parameter matching table composed of multiple groups of data arranged in descending order of matching degree. Each group of data parameters in the parameter matching table includes noise type, time scaling parameter, and slope transformation parameter, where the noise type is one of the k types of ideal noise types.

[0011] According to the above main features, using the trained intelligent neural network model to automatically identify the actual noise includes obtaining the actual noise of the external environment, performing MFCC feature extraction after normalization, and inputting the extracted MFCC parameters to the trained intelligent neural network model for noise recognition calculation. The intelligent neural network model outputs a parameter matching table of the actual noise arranged in descending order of matching degree. Each group of data parameters in the table includes noise type, time scaling parameter, and slope transformation parameter. According to the parameter matching table of the actual noise, select a group of parameters that is consistent with the ideal noise category selected by the user and has the highest ranking among all types of this category in the parameter matching table as the parameters that most closely match the ideal noise. The most matching parameters include noise type, time scaling parameter, and slope transformation parameter, and at the same time, obtain the amplitude ratio of the actual noise.

[0012] According to the above main features, the adjustment of the ideal noise that matches to generate the masking sound according to the most matching parameters including the noise type, time scaling parameter, slope transformation parameter, and actual noise amplitude ratio includes: loading the corresponding ideal noise according to the matching noise type, after normalizing the loaded ideal noise, first performing time scaling processing according to the matching time scaling parameter to achieve the same rhythm as the actual noise, and then the audio characteristic adjustment filter performing audio characteristic slope transformation on the time-scaled ideal noise according to the matching slope transformation parameter to achieve the audio characteristic slope required for masking noise, and then adjusting the volume of the ideal noise after audio characteristic adjustment according to the actual noise amplitude ratio to generate the required masking sound.

[0013] According to the above main features, the method for masking noise further includes receiving an instruction to enable the masking function through an application of an external electronic device, and selecting a preferred ideal noise source category from the four options of "Automatic", "White Noise", "Pink Noise", and "Brown Noise". Then, according to the ideal noise type covered by the selected ideal noise source category by the user and the list of ideal noise types provided by the system in advance for this category, a set of parameters that meet the requirements of the ideal noise category selected by the user and are ranked highest among all types in this category in the parameter matching table is selected as the parameters most matching the ideal noise. The most matching parameters include the noise type, time scaling parameter, and slope transformation parameter, and output is used to generate the masking sound.

[0014] According to the above main features, the method for masking noise further includes receiving a volume adjustment instruction input by the user using the volume adjustment button of the external electronic device, and adjusting the volume of the masking sound according to the instruction input by the user to obtain a more satisfactory auditory experience.

[0015] To achieve the above object, the present invention provides a system for masking noise, and the system for masking noise includes:

[0016] A noise sampling unit for collecting external environmental noise;

[0017] A first amplitude normalization processing unit for normalizing the amplitude of the collected external environmental noise;

[0018] An MFCC feature extraction unit for extracting the features of the external environmental noise;

[0019] An NN neural network model for performing noise recognition calculations and outputting a parameter matching table of the actual noise, where each set of data parameters in the table includes the noise type, time scaling parameter, and slope transformation parameter;

[0020] An amplitude ratio calculation unit for calculating the amplitude ratio of the external environmental noise;

[0021] A selector, which has a list of ideal noise source types pre-provided by the system. The list of ideal noise source types includes three categories (white noise, pink noise, and brown noise), a total of k ideal noise sources. The selector selects a set of parameters that meet the requirements of the ideal noise category selected by the user and are ranked highest among all types in the parameter matching table for this category in the parameter matching table as the most matching parameters according to the ideal noise source category selected by the user and the ideal noise types covered by this category in the list of ideal noise source types pre-provided by the system. The most matching parameters include noise type, time scaling parameter, and slope transformation parameter;

[0022] An ideal noise loading unit, which loads the corresponding ideal noise according to the selected most matching noise type;

[0023] A second amplitude normalization processing unit, which performs normalization processing on the loaded ideal noise;

[0024] A time scaling processing unit, which performs time scaling processing on the normalized ideal noise according to the selected most matching time scaling parameter to achieve the same rhythm as the actual noise;

[0025] An audio characteristic adjustment filter, which performs audio characteristic slope transformation on the time-scaled ideal noise according to the selected most matching slope transformation parameter to achieve the audio characteristic slope required for masking noise;

[0026] A volume adjustment unit, which adjusts the volume of the ideal noise after audio characteristic adjustment according to the amplitude ratio output by the amplitude ratio calculation unit to generate the required masking sound.

[0027] According to the above main features, the noise masking system further includes an external volume adjustment unit, which is connected to the volume adjustment unit so that the user can adjust the volume of the masking sound to obtain a more satisfactory auditory experience.

[0028] According to the above main features, the noise masking system further includes an APP setting unit, which is connected to the selector so that the user can choose to turn on the masking function, select a favorite ideal noise source category from the four options of "automatic", "white noise", "pink noise", and "brown noise". The selector selects a set of parameters that meet the requirements of the ideal noise category selected by the user and are ranked highest among all types in the parameter matching table for this category in the parameter matching table of the actual noise as the most matching parameters according to the ideal noise source category selected by the user and the ideal noise types covered by this category in the list of ideal noise source types pre-provided by the system. The most matching parameters include noise type, time scaling parameter, and slope transformation parameter, and outputs for generating the masking sound.

[0029] Compared with the prior art, the present invention adjusts the audio characteristics by time-scaling an ideal noise source and transforming the audio spectrum slope, so as to obtain more training data with a small number of ideal noise sources. And by time-scaling the most matching ideal noise source selected to obtain the same rhythm as the actual peripheral noise, and by adjusting the spectrum slope and volume of the ideal noise signal processed by time-scaling, thereby improving the masking threshold of the ideal noise, so as to generate a masking sound that can mask the actual noise and meet the auditory perception. The masking sound is played to achieve the masking effect on the actual noise, so as to achieve the purpose of masking the noise.

Description of the Drawings

[0030] Figure 1 It is a schematic diagram of the training principle of the neural network model applied in the method for masking noise of the present invention.

[0031] Figure 2 It is a schematic diagram of the automatic matching principle of the ideal noise in the method for masking noise of the present invention.

[0032] Figure 3 It is a schematic diagram of the automatic generation principle of the masking sound in the method for masking noise of the present invention.

[0033] Figure 4 It is a schematic flow chart of the method for masking noise of the present invention.

[0034] Figure 5 It is a schematic diagram of the composition block of the system for masking noise of the present invention.

Detailed Embodiments

[0035] Focusing on the user's need to obtain an auditory pleasure without being affected by noise in a noise environment that may be harmful to human health and cannot be eliminated by noise reduction, so as to keep them focused, calm or relaxed, the present invention proposes a method based on the principle of acoustic "masking effect", using white noise, pink noise and brown noise sources that have the effects of helping sleep, relaxation and masking noise on the human body and making people feel physically and mentally relaxed as "ideal noise" to mask harmful noise.

[0036] When people listen to a sound in a quiet environment, even if the sound is very low, it can be heard, indicating that the hearing threshold of the human ear for this sound can be very low. However, when listening to a sound, if there is another sound (masking sound) at the same time, it will affect the hearing effect of the sound being listened to by the human ear. At this time, the hearing threshold for the sound being listened to will increase. This phenomenon that the auditory sensitivity of the human ear to other sounds is reduced due to the existence of a certain sound is called the "masking effect". Using one sound to mask another sound, its effect depends on the magnitudes of these two sounds in different corresponding frequency bands. If two sounds exist at the same time and the masking sound is stronger in the same frequency band, the generated masking effect is the largest and the masking effect is the best.

[0037] The minimum sound stimulus that can just cause an auditory response in the human ear is called the auditory threshold. The masking value of one sound to another is defined as: due to the presence of the masking sound, the number of decibels (masking amount) by which the auditory threshold of the masked sound must be increased. The increased auditory threshold is called the masking threshold.

[0038] White noise has a uniform energy distribution across the entire frequency range. It can be used to relieve tinnitus, soothe newborns, and is widely used for insomnia. The hissing sound when a TV has no signal and the sound of waves hitting rocks both belong to white noise.

[0039] Pink noise has an energy distribution in the mid - low frequencies and a low energy distribution in the high - frequency region. Research shows that pink noise is consistent with the brain wave rhythm during sleep, which helps to extend the deep sleep time and improve sleep quality. At the same time, there is also research indicating that pink noise can play a role in desensitizing hyperacusis. The sound of raindrops falling on the sidewalk, the rustling of leaves in the breeze, and the sound of a waterfall cascading into a valley all belong to pink noise.

[0040] The power of the frequency components of Brown noise is mainly concentrated in the mid - low frequencies. This noise is deeper and has more low - frequency components than pink noise, and is a "rumbling" sound, such as the operation of a dryer or the deep sound of tides.

[0041] The present invention uses time - scaling technology and audio characteristic adjustment technology (i.e., audio characteristic slope transformation technology) to achieve the masking of harmful noise. The main idea is to first generate more ideal noises by time - scaling adjustment of a small amount of ideal noises, and then generate more ideal noises with an arbitrary amplitude spectrum slope by transforming the audio characteristic slope of the ideal noises, so as to obtain a large amount of training data required for training a neural network model with a small amount of ideal noise data, thereby training an intelligent neural network model with the ability to automatically identify noise; then use the trained intelligent neural network model to automatically identify the actual noise, and obtain the type of ideal noise, time - scaling parameters, and slope - transformation parameters that best match the actual noise; then adjust the matching ideal noise according to the type of ideal noise, time - scaling parameters, and slope - transformation parameters that best match, including two aspects. The first aspect is to perform time - scaling adjustment on the matching ideal noise to achieve the same rhythm as the actual noise; the second aspect is to perform audio characteristic slope transformation and volume adjustment on the matching ideal noise after time - scaling adjustment to increase the masking threshold of the ideal noise, thereby generating a masking sound that can mask the actual noise and meet the auditory perception. The generated masking sound is played through playback devices such as headphones to achieve the masking effect on the actual noise and achieve the purpose of masking the noise.

[0042] Please refer to Figure 1As shown in the figure, it is the schematic diagram of the training principle of the neural network model applied in the method for masking noise of the present invention. Its principle is to first provide a total of k types of three ideal noise sources (white noise, pink noise, and brown noise), as specifically shown in the following table.

[0043]

[0044] The three types of ideal noise sources described above may include air conditioner noise, fan noise, TV no-signal noise, radio no-signal noise, tidal sound, ocean wave sound, stream sound, rain sound, wind blowing through leaves sound, insect chirping sound, waterfall sound, heavy rain sound, etc.

[0045] After that, each ideal noise is normalized, and then each normalized ideal noise is respectively subjected to m types of time scaling processing to obtain k*m types of ideal noises. Then, each time-scaled ideal noise is subjected to n types of audio characteristic slope transformations, and finally k*m*n types of ideal noises are obtained; MFCC feature extraction is performed on the k*m*n types of ideal noises respectively, and the MFCC feature parameters of the k*m*n types of ideal noises obtained are used as the training data of the neural network model to train the neural network model. The trained intelligent neural network model can perform recognition calculations on the input noise and form a parameter matching table composed of multiple groups of data arranged in descending order of matching degree. Each group of data parameters in the parameter matching table includes noise type, time scaling parameter, and slope transformation parameter, where the noise type is one of the k types of ideal noise types.

[0046] Please refer to Figure 2 As shown in the figure, it is the schematic diagram of the automatic matching of ideal noise in the method for masking noise of the present invention. First, the actual noise in the application environment is obtained. After the actual noise is normalized, MFCC feature extraction is performed. The extracted MFCC parameters are input to the trained neural network model for noise recognition calculation. The intelligent neural network model outputs a parameter matching table of the actual noise arranged in descending order of matching degree. Each group of data parameters in the table includes: noise type KoN, time scaling parameter TS, and slope transformation parameter Rate;

[0047] The parameter matching table of the actual noise, where each set of data parameters in the table includes the noise type KoN, the time scaling parameter TS, and the slope transformation parameter Rate, is input to the selector for selecting the ideal noise type and parameters. Specifically, the user selects to turn on the masking function in the application (APP) of an electronic device (such as a mobile phone or a tablet computer), selects a preferred ideal noise source category from the four options of "Automatic", "White Noise", "Pink Noise", and "Brown Noise". The selector selects a set of parameters that meet the requirements of the ideal noise category selected by the user and has the highest ranking among all types in the category in the parameter matching table as the parameters most matching the ideal noise according to the ideal noise type covered by the selected ideal noise source category and the list of ideal noise types pre-provided by the system. The most matching parameters include the noise type KoN', the time scaling parameter TS', and the slope transformation parameter Rate', which are used to generate the masking sound.

[0048] At the same time, the system calculates the amplitude ratio MR based on the actual noise audio and its normalized audio to adjust the volume of the masking sound. The amplitude ratio calculation formula: MR = actual amplitude of the noise / normalized amplitude of the noise.

[0049] Please refer to Figure 3 As shown, it is the schematic diagram of the automatic generation of the masking sound in the method for masking noise of the present invention. The system loads the selected ideal noise according to the most matching noise type KoN', normalizes the loaded ideal noise, first performs time scaling processing according to the most matching time scaling parameter TS' selected to achieve the same rhythm as the actual noise, and then the audio characteristic adjustment filter performs audio characteristic slope transformation on the time-scaled ideal noise according to the most matching slope transformation parameter Rate' selected to achieve the audio characteristic slope required for masking the noise. Then, the volume of the ideal noise after audio characteristic adjustment is adjusted according to the amplitude ratio MR, thereby generating the required masking sound. If necessary, the user can also adjust the volume of the masking sound by adjusting the volume adjustment button on the electronic device (such as a mobile phone, a tablet computer, or headphones) to obtain a more satisfactory auditory experience.

[0050] Please refer to Figure 4 As shown, it is the schematic flow diagram of the method for masking noise of the present invention. The method includes the following steps:

[0051] Train the intelligent neural network model with multiple ideal noises to obtain a trained intelligent neural network model. The trained intelligent neural network model identifies the input noise and forms a parameter matching table composed of multiple sets of data arranged in descending order of matching degree. Each set of data parameters in the parameter matching table includes the noise type, the time scaling parameter, and the slope transformation parameter;

[0052] Use the trained intelligent neural network model to identify the actual noise, and obtain a parameter matching table of the actual noise arranged from high to low according to the matching degree. Each set of data parameters in the table includes the noise type, time scaling parameter, and slope transformation parameter. At the same time, obtain the amplitude ratio of the actual noise;

[0053] According to the parameter matching table, select a set of parameters that is consistent with the ideal noise category selected by the user and ranks highest among all types of this category in the parameter matching table as the parameters that best match the ideal noise. The parameters that best match include the noise type, time scaling parameter, and slope transformation parameter;

[0054] Adjust the matching ideal noise according to the noise type, time scaling parameter, slope transformation parameter that best match, and the amplitude ratio of the actual noise to generate a masking sound.

[0055] Please refer to Figure 5 As shown, it is a schematic block diagram of the composition of the system for masking noise implementing the present invention. The system for masking noise implementing the present invention includes:

[0056] A noise sampling unit for collecting external environmental noise;

[0057] A first amplitude normalization processing unit for normalizing the amplitude of the collected external environmental noise;

[0058] An MFCC feature extraction unit for extracting the features of the external environmental noise;

[0059] An NN neural network model for performing noise recognition calculations and outputting a parameter matching table of the actual noise. Each set of data parameters in the table includes the noise type, time scaling parameter, and slope transformation parameter;

[0060] An amplitude ratio calculation unit for calculating the amplitude ratio of the external environmental noise;

[0061] A selector is provided with a list of ideal noise source types pre-provided by the system. The list of ideal noise source types includes three categories (white noise, pink noise, and brown noise) with a total of k ideal noise sources. The selector selects a set of parameters that meet the requirements of the ideal noise category selected by the user and ranks highest among all types of this category in the parameter matching table as the parameters that best match according to the ideal noise source category selected by the user and the ideal noise types covered by this category in the list of ideal noise source types pre-provided by the system. The parameters that best match include the noise type, time scaling parameter, and slope transformation parameter;

[0062] An ideal noise loading unit for loading the corresponding ideal noise according to the selected noise type that best matches;

[0063] The second amplitude normalization processing unit normalizes the loaded ideal noise;

[0064] The time scaling processing unit performs time scaling processing on the normalized ideal noise according to the most matched time scaling parameter selected, so as to achieve the same rhythm as the actual noise;

[0065] The audio characteristic adjustment filter performs audio characteristic slope transformation on the time-scaled ideal noise according to the most matched slope transformation parameter selected, so as to achieve the audio characteristic slope required for masking noise;

[0066] The volume adjustment unit adjusts the volume of the ideal noise after audio characteristic adjustment according to the amplitude ratio output by the amplitude ratio calculation unit to generate the required masking sound.

[0067] According to the above main features, the noise masking system further includes an external volume adjustment unit connected to the volume adjustment unit, so that the user can adjust the volume of the masking sound to obtain a more satisfactory auditory experience.

[0068] According to the above main features, the noise masking system further includes an APP setting unit connected to the selector, so that the user can choose to turn on the masking function, select a favorite ideal noise source category from the four options of "automatic", "white noise", "pink noise", and "brown noise", and the selector selects a set of parameters that meet the requirements of the ideal noise category selected by the user and are ranked highest among all types in the category in the parameter matching table of the actual noise as the most matched parameters according to the ideal noise source category selected by the user and the ideal noise types covered by this category in the list of ideal noise sources provided in advance by the system. The most matched parameters include noise type, time scaling parameter, and slope transformation parameter, and are output to generate the masking sound.

[0069] In specific implementation, the first and second amplitude normalization processing units can be set as the same entity.

[0070] In addition, for the relevant descriptions of audio time scaling technology, reference can be made to the literature titled "Improved phase vocoder time-scale modification of audio" published in IEEE Transactions on Speech and Audio Processing (Volume: 7, Issue: 3, May 1999) in 1999, as well as U.S. Patents 5,717,768 and 6,868,377. The audio spectrum slope transformation technology is a technology that adjusts the audio frequency response characteristics by performing audio characteristic slope transformation based on slope transformation parameters. The slope refers to the magnitude of the octave attenuation of the noise frequency response characteristics. For example, the "slope" of white noise is 0 dB, the "slope" of pink noise is 3 dB, and the slope of brown noise is 6 dB. In order to obtain the best masking noise, the neural network model selects the most matching octave attenuation. An ideal sound source has its own octave attenuation. As mentioned above, the octave attenuation of pink noise is 3 dB, and that of brown noise is 6 dB, etc. The difference between the most matching octave attenuation selected by the neural network model and the octave attenuation inherent in the ideal sound source is called the "slope transformation parameter". Based on multiple filters, the frequency response characteristics required to implement the "slope transformation parameter" are segmented and fitted to generate the best masking noise. For the relevant prior art, reference can be made to Chinese Patent Applications 201310377482.8, 201710991542.3, and 201811190929.X, which will not be elaborated here.

[0071] Compared with the prior art, the present invention adjusts the audio characteristics by performing time scaling on the ideal noise source and transforming the audio spectrum slope, so as to obtain more training data with a small number of ideal noise sources, and perform time scaling on the most matching ideal noise source selected to obtain the same rhythm as the actual peripheral noise, and perform spectrum slope adjustment and volume adjustment on the ideal noise signal after time scaling processing, thereby improving the masking threshold of the ideal noise, so as to generate a masking sound that can mask the actual noise and meet the auditory perception. The masking sound is played to achieve the masking effect on the actual noise, so as to achieve the purpose of masking the noise.

[0072] It can be understood that for those of ordinary skill in the art, equivalent substitutions or changes can be made according to the technical solution of the present invention and its inventive concept, and all such changes or substitutions should fall within the protection scope of the claims appended to the present invention.

Claims

1. A method for masking noise, characterized in that The method for masking noise includes the following steps: Training an intelligent neural network model with a variety of ideal noises. The trained intelligent neural network model identifies the input noises and forms a parameter matching table composed of multiple groups of data arranged in descending order of matching degree. Each group of data parameters in the parameter matching table includes noise type, time scaling parameter, and slope transformation parameter; Using the trained intelligent neural network model to identify the actual noise, obtaining a parameter matching table of the actual noise arranged in descending order of matching degree. Each group of data parameters in the table includes noise type, time scaling parameter, and slope transformation parameter, and at the same time obtaining the amplitude ratio of the actual noise; According to the parameter matching table, select a group of parameters that are consistent with the ideal noise category selected by the user and have the highest ranking among all types of this category in the parameter matching table as the parameters most matching the ideal noise. The most matching parameters include noise type, time scaling parameter, and slope transformation parameter; Adjust the matching ideal noise according to the most matching noise type, time scaling parameter, slope transformation parameter, and the amplitude ratio of the actual noise to generate a masking sound.

2. The method for masking noise according to claim 1, wherein: The step of training the intelligent neural network model with a variety of ideal noises includes providing a total of k types of ideal noise sources of white noise, pink noise, and Brown noise. First, normalize each type of ideal noise, then perform m types of time scaling processing on each normalized ideal noise to obtain k*m types of ideal noises, and then perform n types of audio characteristic slope transformation on each time-scaled ideal noise to finally obtain k*m*n types of ideal noises; perform MFCC feature extraction on the k*m*n types of ideal noises respectively, and use the MFCC feature parameters of the k*m*n types of ideal noises obtained as the training data of the neural network model to train the neural network model. The trained intelligent neural network model performs identification and calculation on the input noises and forms a parameter matching table composed of multiple groups of data arranged in descending order of matching degree. Each group of data parameters in the parameter matching table includes noise type, time scaling parameter, and slope transformation parameter, where the noise type is one of the k types of ideal noise types.

3. The method for masking noise according to claim 2, wherein: Automatically identifying the actual noise using the trained intelligent neural network model includes obtaining the actual noise of the external environment, performing MFCC feature extraction after normalization, inputting the extracted MFCC parameters into the trained intelligent neural network model for noise identification and calculation. The intelligent neural network model outputs a parameter matching table of the actual noise arranged in descending order of matching degree. Each group of data parameters in the table includes noise type, time scaling parameter, and slope transformation parameter. According to the parameter matching table of the actual noise, select a group of parameters that are consistent with the ideal noise category selected by the user and have the highest ranking among all types of this category in the parameter matching table as the parameters most matching the ideal noise. The most matching parameters include noise type, time scaling parameter, and slope transformation parameter, and at the same time obtain the amplitude ratio of the actual noise.

4. The method for masking noise according to claim 3, wherein: Adjusting the matched ideal noise according to the most matched parameters including noise type, time scaling parameter, slope transformation parameter, and actual noise amplitude ratio to generate a masking sound includes: loading the corresponding ideal noise according to the matched noise type, normalizing the loaded ideal noise, first performing time scaling processing according to the matched time scaling parameter to achieve the same rhythm as the actual noise, then performing audio characteristic slope transformation on the time-scaled ideal noise by an audio characteristic adjustment filter according to the matched slope transformation parameter to achieve the audio characteristic slope required for masking noise, and then adjusting the volume of the ideal noise after audio characteristic adjustment according to the actual noise amplitude ratio to generate the required masking sound.

5. The method for masking noise according to claim 4, wherein: The method for masking noise further includes receiving an instruction to enable the masking function through an application of an external electronic device, and selecting a preferred ideal noise source category from four options: "Automatic", "White Noise", "Pink Noise", and "Brown Noise". Then, according to the ideal noise type covered by the selected ideal noise source category in the list of ideal noise sources pre-provided by the system, a set of parameters that meet the requirements of the selected ideal noise category by the user and are ranked highest among all types in this category in the parameter matching table is selected as the most matched parameters for the ideal noise. The most matched parameters include noise type, time scaling parameter, and slope transformation parameter, and are output for generating the masking sound.

6. The method for masking noise according to claim 5, wherein: The method for masking noise further includes receiving a volume adjustment instruction input by the user using the volume adjustment button of the external electronic device, and adjusting the volume of the masking sound according to the instruction input by the user to obtain a more satisfactory auditory experience.

7. A noise masking system, characterized in that The system for masking noise includes: A noise sampling unit for collecting external environmental noise; A first amplitude normalization processing unit for normalizing the amplitude of the collected external environmental noise; An MFCC feature extraction unit for extracting the features of the external environmental noise; An NN neural network model for performing noise recognition calculations and outputting a parameter matching table of the actual noise, where each set of data parameters in the table includes noise type, time scaling parameter, and slope transformation parameter; An amplitude ratio calculation unit for calculating the amplitude ratio of the external environmental noise; A selector with a list of ideal noise sources pre-provided by the system. The list of ideal noise sources includes k ideal noise sources in three categories: white noise, pink noise, and brown noise. The selector selects a set of parameters that meet the requirements of the selected ideal noise category by the user and are ranked highest among all types in this category in the parameter matching table as the most matched parameters according to the selected ideal noise source category by the user and the ideal noise types covered by this category in the list of ideal noise sources pre-provided by the system. The most matched parameters include noise type, time scaling parameter, and slope transformation parameter; An ideal noise loading unit for loading the corresponding ideal noise according to the selected most matched noise type; A second amplitude normalization processing unit for normalizing the loaded ideal noise; A time scaling processing unit that performs time scaling processing on the normalized ideal noise according to the most matched time scaling parameter selected to achieve the same rhythm as the actual noise; An audio characteristic adjustment filter that performs audio characteristic slope transformation on the time-scaled ideal noise according to the most matched slope transformation parameter selected to achieve the audio characteristic slope required for masking noise; A volume adjustment unit that adjusts the volume of the ideal noise after audio characteristic adjustment according to the amplitude ratio output by the amplitude ratio calculation unit to generate the required masking sound.

8. The noise masking system according to claim 7, wherein: The noise masking system further includes an external volume adjustment unit connected to the volume adjustment unit so that the user can adjust the volume of the masking sound to obtain a more satisfactory auditory experience.

9. The noise masking system according to claim 8, wherein: The noise masking system further includes an APP setting unit connected to the selector so that the user can choose to turn on the masking function, select a preferred ideal noise source category from the four options of "automatic", "white noise", "pink noise", and "brown noise". The selector selects a set of parameters that meet the requirements of the ideal noise category selected by the user and has the highest ranking among all the types in the parameter matching table of the actual noise as the most matched parameters according to the ideal noise source type list provided in advance by the system for the ideal noise source category selected by the user. The most matched parameters include the noise type, time scaling parameter, and slope transformation parameter, and are output to generate the masking sound.

Images