White noise adaptive recommendation method and device based on electroencephalogram signal feedback

Abstract

The application relates to a white noise adaptive recommendation method and device based on electroencephalogram signal feedback. The method comprises the following steps: controlling an audio playing device to play different types of white noises, wherein the volume levels of the different types of white noises are the same; collecting each type of electroencephalogram signal under the stimulation of the different types of white noises, and determining at least one target type of white noise based on each type of electroencephalogram signal; for each target type of white noise, controlling the audio playing device to play the white noise with different volume levels; collecting volume electroencephalogram signals under the white noises with the volume levels, and determining the white noise under a target volume level based on the volume electroencephalogram signals; and determining the white noise with the target volume level of the target type as the starting historical preference white noise based on the determined at least one target volume level of white noise and at least one target type of white noise. The method can improve the accuracy of the white noise.

Description

White Noise Adaptive Recommendation Method and Device Based on EEG Signal Feedback Technical Field

[0001] This application relates to the field of computer technology, and in particular to a white noise adaptive recommendation method and apparatus based on EEG signal feedback. Background Technology

[0002] During EEG signal acquisition, ambient sound is a significant factor affecting signal quality. Traditional EEG acquisition is often conducted in completely silent environments or scenes with environmental interference. These environments may cause participants' attention to drift or drowsiness, affecting signal stability. Appropriate amounts of background white noise can shield against environmental noise, improve concentration, and thus improve EEG data quality. White noise refers to a sound whose frequency is uniform within the audible range, creating a masking effect that blocks out and ignores subtle changes in external sounds. Previous white noise studies have demonstrated its ability to improve attention, work efficiency, and sleep quality. Recorded or synthesized sounds that perceptually approximate white noise characteristics, but with specific life or natural scene features, can help participants enter a focused state more quickly, improving comfort during EEG acquisition. However, the optimization effects of different types of white noise and different volume levels vary significantly among individuals.

[0003] Traditional technologies typically employ fixed, single ambient sound conditions or even no ambient sound, lacking an individualized optimization process. This results in unstable EEG signal quality and a suboptimal signal-to-noise ratio, affecting subsequent data analysis and brain-computer interface performance. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and device for adaptive white noise recommendation based on EEG signal feedback that can improve the accuracy of white noise and address the aforementioned technical problems.

[0005] In a first aspect, this application provides a white noise adaptive recommendation method based on EEG signal feedback, the method further comprising:

[0006] The audio playback device is controlled to play different types of white noise, and the volume levels of the different types of white noise are the same;

[0007] Collect various types of electroencephalogram (EEG) signals under stimulation by different types of white noise, and determine at least one target type of white noise based on the EEG signals of each type.

[0008] For each of the target types of white noise, the audio playback device is controlled to play white noise at different volume levels;

[0009] Collect volume EEG signals at each of the specified volume levels under the white noise, and determine the white noise at the target volume level based on the volume EEG signals;

[0010] Based on the determined white noise at at least one target volume level and white noise of at least one target type, white noise at the target volume level under the target type is determined as the starting historical preferred white noise.

[0011] In some alternative embodiments, the method further includes:

[0012] The audio playback device is controlled to send identification information to the EEG acquisition device. The identification information carries the sound type and time identifier. The identification information is used to identify the EEG signals acquired by the EEG acquisition device when playing each white noise.

[0013] In some optional embodiments, the target type of white noise and the white noise at the target volume level are determined by the following methods:

[0014] For each current white noise, feature extraction is performed on the EEG signal to obtain EEG signal features, wherein when the target type of white noise is determined, the current white noise is white noise of the type, and the EEG signal includes the type of EEG signal; when the target volume level of white noise is determined, the current white noise is the white noise at each volume level, and the EEG signal is the volume EEG signal.

[0015] The values ​​of each white noise assessment index are determined based on the characteristics of the electroencephalogram (EEG) signals.

[0016] The dynamic weights of each white noise assessment index are determined based on its value.

[0017] The current white noise evaluation result is obtained based on the dynamic weights and the values ​​of each of the white noise evaluation indicators.

[0018] Based on the evaluation results of each of the current white noises, white noise to be screened is determined. If the target type of white noise is determined, the white noise to be screened is white noise of at least one target type. If the target volume level of white noise is determined, the white noise to be screened is white noise of at least one target volume level.

[0019] In some optional embodiments, determining the values ​​of each white noise assessment index based on the EEG signal characteristics includes:

[0020] For each current white noise, based on the EEG signal characteristics, the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise are obtained.

[0021] Based on the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise, the initial sound sensitivity index value is obtained.

[0022] The target sound sensitivity index value is obtained by normalizing the initial sound sensitivity index value.

[0023] For each current white noise, the target fractional power ratio of different waves is determined based on the characteristics of the EEG signal, and the initial attention index value is determined based on the target fractional power ratio of different waves.

[0024] The initial attention index value of the current white noise is normalized to obtain the normalized attention index value;

[0025] Candidate attention index values ​​are obtained based on the normalized attention index values ​​of each current white noise, and the candidate attention index values ​​are normalized to obtain the target attention index value.

[0026] In some optional embodiments, determining the dynamic weight of each white noise evaluation index based on its value includes:

[0027] Based on the white noise evaluation index values, the signal-to-noise ratio of each index is obtained;

[0028] The dynamic weights of each white noise evaluation index are determined based on the signal-to-noise ratio of each index.

[0029] Secondly, this application also provides a target white noise adaptive recommendation method, the method comprising:

[0030] When entering a new environment, extract the sound features of the new environment;

[0031] Historical preference white noise is obtained based on the current task scenario, and the historical preference white noise is determined based on the white noise adaptive recommendation method based on EEG signal feedback described above.

[0032] Based on the historical preferred white noise and the sound characteristics of the new environment, the target sound characteristics of the new environment are obtained;

[0033] The target white noise corresponding to the new environment is determined based on the target sound characteristics.

[0034] In some alternative embodiments, the extraction of the acoustic features of the new environment includes:

[0035] Extract the sound signals of the new environment;

[0036] The sound signal is preprocessed, and the preprocessed sound signal is feature extracted to obtain sound features, the sound features including at least one of environmental entropy value, rhythm intensity and main frequency;

[0037] The process of obtaining the target sound features of the new environment based on the historical preferred white noise and the sound features of the new environment includes:

[0038] Obtain the sound features of the old environment corresponding to the historical preferred white noise;

[0039] Based on the sound characteristics of the old environment and the historical preferred white noise, differential sound characteristics are obtained;

[0040] Based on the difference in sound features and the sound features of the new environment, the target sound features of the new environment are obtained;

[0041] The step of determining the target white noise corresponding to the new environment based on the target sound features includes:

[0042] Determine the pre-calculated sound characteristics of each white noise in the white noise library;

[0043] Calculate the pre-calculated sound features of each white noise in the white noise library and the Euclidean weighted distance between them and the sound features of the new environment;

[0044] Select the target number of white noises whose Euclidean weighted distance meets the requirements as candidate white noises;

[0045] Each of the candidate white noises is played in sequence, and the target white noise is determined based on the interactive instructions.

[0046] In some alternative embodiments, the method further includes:

[0047] Receive artifact ratio and signal-to-noise ratio data output by the EEG device, and determine the data quality based on the artifact ratio and signal-to-noise ratio data;

[0048] If the correlation between the changes in the sound features of the new environment and the data quality is greater than a correlation threshold, the sound features of the new environment are adjusted to obtain a new target white noise, which is then stored as a new historical preferred white noise.

[0049] Thirdly, this application also provides a white noise adaptive recommendation device based on EEG signal feedback, the device comprising:

[0050] A white noise playback module is used to control an audio playback device to play different types of white noise, and the volume levels of the different types of white noise are the same.

[0051] A white noise type determination module is used to collect various types of EEG signals under stimulation by various types of white noise, and to determine at least one target type of white noise based on the various types of EEG signals.

[0052] A volume white noise playback module is used to control the audio playback device to play white noise at different volume levels for each of the target types of white noise;

[0053] A volume white noise determination module is used to collect volume EEG signals under the white noise at each of the volume levels, and determine the white noise at the target volume level based on the volume EEG signals.

[0054] A preference determination module is used to determine, based on at least one target volume level of white noise and at least one target type of white noise, a target volume level of white noise as the starting historical preference white noise.

[0055] Fourthly, this application also provides a target white noise adaptive recommendation device, the device comprising:

[0056] An extraction module is used to extract the sound features of a new environment when entering a new environment;

[0057] The historical preference acquisition module is used to acquire historical preference white noise based on the current task scenario. The historical preference white noise is determined based on the white noise adaptive recommendation device based on EEG signal feedback in any of the above embodiments.

[0058] The sound feature determination module is used to obtain the target sound features of the new environment based on the historical preferred white noise and the sound features of the new environment;

[0059] The target white noise determination module is used to determine the target white noise corresponding to the new environment based on the target sound features.

[0060] The aforementioned adaptive white noise recommendation method and apparatus based on EEG signal feedback involves controlling an audio playback device to play different types of white noise, with each type of white noise having the same volume level; acquiring various types of EEG signals under the stimulation of the different types of white noise, and determining at least one target type of white noise based on the EEG signals of each type; controlling the audio playback device to play white noise at different volume levels for each target type of white noise; acquiring volume EEG signals under the white noise at each volume level, and determining white noise at a target volume level based on the volume EEG signals; and determining white noise at a target volume level based on the determined white noise at at least one target volume level and at least one target type of white noise as the initial historical preferred white noise. This method improves the accuracy of white noise determination based on individual differences, thereby enhancing the quality of subsequent EEG signals. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 is a flowchart of a white noise adaptive recommendation method based on EEG signal feedback in one embodiment;

[0063] Figure 2 is a schematic diagram of the audio playback device during the sound type test in Phase 1 of one embodiment;

[0064] Figure 3 is a schematic diagram of the audio playback device during the sound intensity test in stage two of an embodiment;

[0065] Figure 4 is a flowchart of the method for determining the target type of white noise and the target volume level of white noise in one embodiment;

[0066] Figure 5 is a flowchart of the feature extraction steps in one embodiment;

[0067] Figure 6 is a block diagram of a white noise adaptive recommendation method based on EEG signal feedback in one embodiment;

[0068] Figure 7 is a flowchart of a white noise adaptive recommendation method based on EEG signal feedback in another embodiment;

[0069] Figure 8 is a flowchart illustrating the target white noise adaptive recommendation method in one embodiment;

[0070] Figure 9 is a structural block diagram of a white noise adaptive recommendation device based on EEG signal feedback in one embodiment;

[0071] Figure 10 is a structural block diagram of a target white noise adaptive recommendation device in one embodiment;

[0072] Figure 11 is an internal structure diagram of a computer device in one embodiment. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0074] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0075] In one embodiment, as shown in Figure 1, a white noise adaptive recommendation method based on EEG signal feedback is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0076] S102: Controls the audio playback device to play different types of white noise, with each type of white noise having the same volume level.

[0077] As shown in Figure 2, which is a schematic diagram of the audio playback device during the sound type test in one embodiment, the audio playback device sequentially presents different types of white noise (such as natural white noise, weather white noise, mechanical white noise, etc.) at a consistent volume (e.g., 40dB; in other embodiments, the volume can also be other values, which are not specifically limited here). Furthermore, it should be noted that each condition includes multiple rounds, with fixed or random intervals between rounds to avoid auditory adaptation.

[0078] The playback control module can call up pre-stored high-fidelity audio files of various categories, including natural sounds (waves, rainforest, forest wind, streams), weather sounds (moderate rain, drizzle, snowfall, wind howling), mechanical sounds (fans, air conditioners, low-speed engine operation), and artificially synthesized broadband noise or pink noise, and output them to speakers or closed-back headphones at the set volume and phase without distortion through a professional audio interface.

[0079] S104: Collect various types of EEG signals under different types of white noise stimulation, and determine at least one target type of white noise based on the various types of EEG signals.

[0080] In this application, an EEG signal acquisition module can be used to acquire various types of EEG signals under different types of white noise stimulation, and at least one target type of white noise can be determined based on the type of EEG signal.

[0081] The EEG acquisition module utilizes a multi-channel amplifier and electrode cap combination. The electrode layout conforms to the international 10-40 system standard, and additional channels can be added to cover relevant areas of the auditory cortex according to experimental needs. The sampling rate is ≥500Hz, and the amplifier has an input impedance higher than 1GΩ and a common-mode rejection ratio >100dB to improve the signal-to-noise ratio. During acquisition, high-resolution EEG waveforms are written to the buffer via a low-latency data stream interface, along with a timestamp of the sound playback trigger signal. After receiving the raw EEG data, the data processing module first performs bandpass filtering, typically in the frequency range of 1-40Hz, to remove low-frequency drift and high-frequency electromyography interference. The second step uses independent component analysis (ICA) to separate eye movement, blink, and electrocardiogram components, and removes these artifact components to retain waveform components representing the true brain response. Subsequently, segments from -500ms to 500ms are extracted according to the trigger time window, and baseline correction is performed on each segment, eliminating the influence of slow fluctuations by using the average potential before stimulation as a benchmark. During the standardization phase, the module performs Z-score transformation on the amplitude of each segment to eliminate overall amplitude differences across subjects and across conditions, so that subsequent analysis only reflects the changes caused by the stimulus conditions themselves.

[0082] S106: For each type of white noise, control the audio playback device to play white noise at different volume levels.

[0083] Referring to Figure 3, which is a playback diagram of the audio playback device during the sound intensity test in stage two of one embodiment, multiple volume levels are presented sequentially under the optimal noise type selected in the first stage (e.g., 50dB, 40dB, 50dB, 650dB; other levels may also be used in other embodiments, and no specific limitation is made here). It should also be noted that each condition includes multiple rounds, with fixed or random intervals between rounds to avoid auditory adaptation.

[0084] S108: Collect volume EEG signals under white noise at various volume levels, and determine the white noise at the target volume level based on the volume EEG signals.

[0085] In this application, an EEG signal acquisition module can be used to acquire EEG signals of different volumes under the stimulation of white noise at different volume levels, and to determine white noise at at least one target volume level based on the volume EEG signals.

[0086] For ease of understanding, the EEG signal acquisition module can acquire EEG signals under noise stimulation conditions, which can be ERP signals.

[0087] In some optional embodiments, the method further includes: controlling the audio playback device to send identification information to the EEG acquisition device. The identification information carries a sound type and a time marker, and is used to identify the EEG signals acquired by the EEG acquisition device during the playback of each white noise. This synchronizes the stimulus event with the EEG data, including noise type, volume level, and cycle information. Specifically, during playback, the playback control module sends a digital or analog trigger signal to the EEG acquisition device. The trigger signal carries a unique sound category and time marker digital code, thereby ensuring accurate matching between the event marker and the sound stimulus.

[0088] S110: Based on the determined white noise of at least one target volume level and white noise of at least one target type, determine the white noise of the target volume level under the target type as the starting historical preferred white noise.

[0089] Specifically, in this application, white noise of at least one target volume level and white noise of at least one target type are combined to form a final candidate parameter set.

[0090] If the final candidate parameter set contains only one parameter, output the unique optimal combination as the initial historical preference white noise. If the final candidate parameter set contains more than one parameter, output the candidate set and randomly select one of them as the initial historical preference white noise.

[0091] The aforementioned adaptive white noise recommendation method based on EEG signal feedback controls an audio playback device to play different types of white noise at the same volume level. It collects various types of EEG signals under the stimulation of different types of white noise and determines at least one target type of white noise based on these signals. For each target type of white noise, the audio playback device plays white noise at different volume levels. It collects volume EEG signals at each volume level and determines the target volume level of white noise based on these signals. Based on the determined target volume level of white noise and at least one target type of white noise, it determines the target volume level of white noise for the target type as the initial historical preferred white noise. This method determines white noise based on individual differences, improving the accuracy of white noise selection and thus enhancing the quality of subsequent EEG signals.

[0092] In some optional embodiments, referring to Figure 4, which is a flowchart of a method for determining the target type of white noise and the target volume level of white noise in one embodiment, the method for determining the target type of white noise and the target volume level of white noise is the same in this embodiment. Specifically, the target type of white noise and the target volume level of white noise are determined by the following methods:

[0093] S402: For each current white noise, feature extraction is performed on the EEG signal to obtain EEG signal features. Wherein, if the target type of white noise is determined, the current white noise is white noise of type, and the EEG signal includes type EEG signal; if the target volume level of white noise is determined, the current white noise is white noise of each volume level, and the EEG signal is volume EEG signal.

[0094] S404: Determine the values ​​of each white noise assessment index based on EEG signal characteristics.

[0095] S406: Determine the dynamic weight of each white noise evaluation index based on the values ​​of each white noise evaluation index.

[0096] S408: The current white noise assessment result is obtained based on dynamic weights and the values ​​of each white noise assessment index.

[0097] S410: Determine the white noise to be screened based on the evaluation results of each current white noise. If the target type of white noise is determined, the white noise to be screened is white noise of at least one target type. If the target volume level of white noise is determined, the white noise to be screened is white noise of at least one target volume level.

[0098] For ease of understanding, this application uses the method of determining the target type of white noise as an example for illustration:

[0099] All noise types are derived from the condition set. ,in Represents a specific noise type, with a fixed volume. (e.g., 40dB). For each noise type, calculate the normalized ERP amplitude of the stimulus-state ERP at a fixed volume. .

[0100] The specific calculation method includes: calculating the EEG data with a filtering range of 1–40 Hz; selecting a specific time window (-500ms–500ms) as the period of interest based on the time stamp of the noise stimulus and the analysis of the ERP signal and frequency; and calculating the ERP pre-stimulus baseline and performing z-score normalization.

[0101] ERP amplitude calculation formula: (ERP amplitude of target stimulus under certain noise conditions.) Pre-stimulus baseline.

[0102] The calculation of the baseline mean and standard deviation includes: averaging the values ​​from -500ms to 0ms before each trial stimulus. and standard deviation , here and It's time to try my own value.

[0103] Standardize the data:

[0104]

[0105] The EEG signal corresponding to time t of the trial. This represents the standard deviation of the EEG signal at time t corresponding to the trial. This represents the average value of the EEG signal at time t corresponding to the trial. This represents the standardized amplitude. Since the sampled data is actually discrete, the average of the absolute values ​​of the standardized amplitudes is summed. The calculation time window is 100-500 ms after the event.

[0106]

[0107] in The number of sampling points within the time window. The sampling times within the time window are used to obtain the ERP values ​​after z-score baseline normalization. This is called the normalized ERP amplitude.

[0108] Specifically, in the feature extraction stage, the system performs quantitative analysis on event-related potential (ERP) components closely related to cognitive processing and attentional resource allocation, particularly the mean absolute value of waveform amplitude within 100 to 500 milliseconds after the target stimulus. This value reflects the brain's resource investment and neural intensity response to the current sound environment. To obtain effective ERPs, 40 trials of each sound type are played as a baseline. When the SNR (Signal Noise Ratio) of 5 consecutive trials of the same type is ≥3.0, the acquisition of that type can be stopped immediately. After each trial is completed, the SNR = (P500 peak-to-peak value) / (baseline standard deviation) is calculated immediately, and quality control is performed to automatically exclude trials with severe blinking / electromyographic contamination (>±100μV). Based on this, this embodiment introduces the Noise Sensitivity Index (NSI) and Attention Index (ATI) as comprehensive metrics.

[0109] Specifically, the white noise assessment index values ​​are determined based on EEG signal characteristics. These white noise assessment index values ​​include the target sound sensitivity index (NSI) and the normalized attention index (ATI). The calculation formula for the sound sensitivity index is as follows: ,in This represents the mean of the ERP index under a specific target white noise condition. σ represents the mean of this index under other conditions, and σ represents the corresponding standard deviation. A higher NSI value indicates that the sound conditions elicit a stronger neural response and are relatively stable and reliable. The formula for calculating the attention index is... ,in This represents the mean of the attention index under a specific target white noise condition. σ represents the mean of the index under other conditions, and σ represents the corresponding standard deviation.

[0110] In some optional embodiments, referring to Figure 5, which is a flowchart of a feature extraction step in one embodiment, the feature extraction step, namely determining the white noise evaluation index values ​​based on EEG signal features, includes:

[0111] S502: For each current white noise, based on the characteristics of the EEG signal, obtain the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise.

[0112] S504: Based on the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise, the initial sound sensitivity index value is obtained.

[0113] For each type of white noise, calculate the initial sound sensitivity index (NSI) value: subtract the average value of other types from the normalized ERP mean of that type of white noise, and then divide by the square root of the sum of squared variances to measure the difference and stability.

[0114]

[0115] in For condition set The average of the normalized ERP mean values ​​for all white noise types except Ti. The variance of ERP magnitude under the target type. This represents the variance of the ERP magnitude under non-target types.

[0116] S506: Normalize the initial sound sensitivity index value to obtain the target sound sensitivity index value.

[0117] To further dynamically weight the data with ATI, After normalization, the target sound sensitivity index value is obtained. .

[0118]

[0119] in, , It is all types of noise mean It is all types of noise The standard deviation.

[0120] S508: For each current white noise, determine the target power ratio of different waves based on the characteristics of the EEG signal, and determine the initial attention index value based on the target power ratio of different waves.

[0121] The formula for calculating the Initial Attention Index (ATI) is: (ATI of the target stimulus under certain noise conditions.)

[0122]

[0123] in, for The rated power of the wave (4-7Hz), for The power of the theta wave (13-50 Hz) is calculated within a time window of 100-500 ms after the event. An increase in theta waves (4-7 Hz) reflects cognitive load, while a decrease in beta waves (13-50 Hz) indicates suppression of distracting attention. The higher the ratio, the more focused the attention.

[0124] S510: Normalize the initial attention index value of the current white noise to obtain the normalized attention index value.

[0125] Normalization is performed in accordance with NSI to eliminate the extreme differences between trials and standardize the current data.

[0126]

[0127] The ATI mean for all trials of the current noise type. This represents the standard deviation of ATI for all trials of the current noise type. The z-score baseline-normalized ATI values ​​are then obtained. This is called the Normalized Attention Index (ATI), or the Normalized Attention Index value.

[0128] S512: Based on the normalized attention index value of each current white noise, obtain the candidate attention index value, and normalize the candidate attention index value to obtain the target attention index value.

[0129] For each noise type, calculate its normalized ATI index for the stimulus state at a fixed volume. .

[0130] Calculate the ATI value, or candidate attention index value, for each type: subtract the mean of the normalized ATI for that type from the mean of other types, and then divide by the square root of the sum of squared variances to measure the difference and stability.

[0131]

[0132] in The mean, Let ATI be the variance under the target type. This represents the variance of ATI under non-target types.

[0133] To further dynamically weight the candidate attention metric values ​​with ATI, After normalization, the target attention index value is obtained. .

[0134]

[0135] in, , It is all types of noise mean It is all types of noise The standard deviation.

[0136] In some optional embodiments, determining the dynamic weight of each white noise evaluation index based on each white noise evaluation index value includes: obtaining the signal-to-noise ratio of each index based on each white noise evaluation index value; and determining the dynamic weight of each white noise evaluation index based on the signal-to-noise ratio of each index.

[0137] For dynamic weights The calculation will be based on the indicator signal-to-noise ratio (SNR):

[0138]

[0139] The definition of SNR is as follows: , .

[0140] Finally, based on the dynamic weights and the values ​​of each white noise evaluation index, the current white noise evaluation result is obtained, namely:

[0141] right and Dynamic weighted fusion is performed to obtain the fusion score FS. The formula is as follows:

[0142]

[0143] Finally, a global significance test is used to determine whether the overall differences between different types of FS are significant. If significant, pairwise comparisons are performed to identify the set of types that are significantly superior to other types. If the set is empty, multiple types are retained and proceed to the second stage; if there is one set, that type is fixed; if there are multiple sets, multiple types are entered, and the identified types are stored. .

[0144] The NSI and ATI indicators are dynamically weighted based on the signal-to-noise ratio (SNR) during the recording process. The weight is determined by the SNR, thus determining the Fusion Score (FS) under different noise conditions. The FS values ​​of all conditions are tested for significance using one-way ANOVA, supplemented by post-hoc comparisons such as Tukey HSD, to determine whether the difference between the target condition and other conditions reaches statistical significance (p<0.05), ensuring that the screening results are based on scientific evidence rather than random fluctuations.

[0145] For the method of determining the target volume level of white noise, please refer to the method of determining the target type of white noise. This embodiment only provides a brief description:

[0146] All noise types are derived from the condition set. ,in Represents a specific noise type, with a fixed volume. (Results of Phase One). For each decibel level, calculate its... Normalized ERP amplitude of the understimulatory state .

[0147] Calculate the NSI value for each decibel: subtract the average of other types from the normalized ERP mean of that type, and then divide by the square root of the sum of squared variances to measure variability and stability.

[0148]

[0149] in This is the average of the normalized ERP means for all white noise types except Vj in the condition set Cvol. The variance of ERP magnitude under the target type. This represents the variance of the ERP magnitude under non-target types.

[0150] Will Perform normalization.

[0151] Similarly, an ATI value is calculated for each volume level.

[0152] Will Perform normalization. This represents the average AIT (Advanced Interference Time) of all white noise types except Vj in the condition set Cvol. After calculating FS, a global significance test is used to determine whether the overall differences between different volumes are significant. If significant, pairwise comparisons are performed to find the volume set that is significantly better than the others. If the set is empty, multiple volumes are retained and proceed to the second stage; if the set contains one volume, that volume is fixed; if multiple sets contain multiple volumes, all volumes are included. Finally, the found volumes are stored as... .

[0153] In the above embodiment, the operation is divided into two stages: The first stage is the type selection stage. In a quiet background environment (background sound pressure level ≤ 35dB), the system performs pre-screening of categories with a uniform initial volume (e.g., 55dB). Group clustering tests are used to pre-classify white noise into 3-5 categories based on acoustic characteristics (spectral entropy). Representative samples are selected for each category (e.g., rain sounds and ocean waves for the natural category) for short-term testing. Next, in categories with good user test results, further subdivision and self-recommendation of the same type are performed. White noise in that category is played pseudo-randomly, with the playback time and silence interval for each type set according to a fixed scheme. For example, each playback is 5 seconds long, with a 10-second silence interval inserted in between. The entire loop sequence is randomized to avoid sequence effects. This quickly distinguishes between white noise "rich in natural details" and "mechanical monotony," avoiding testing redundant samples with overly similar acoustic characteristics. EEG signals are collected in real-time throughout the playback, and trigger information is recorded synchronously.

[0154] After data processing and NSI / ATI calculations, if statistical analysis shows that a certain type is significantly better than other types, it is selected as the best category for individualized recommendation; if the differences between multiple types are not significant, they are included in the candidate set for future use. The second stage is the volume optimization stage. The playback module only presents the best category of white noise determined in the first stage, but adjusts the volume value before each playback, setting a gradient from 40 to 70 dB, with each increment at 5 dB or smaller. The acquisition, processing, and analysis process is consistent with the first stage, comprehensively determining the volume level that produces the best neural response effect and comfort, and simultaneously collecting the spectral entropy value of the environment through the microphone for pairing and storage.

[0155] For ease of understanding, in one specific embodiment, as shown in Figures 6 and 7, this application is used to screen the optimal type and volume parameters for an individual under different categories of white noise. First, the playback control module 1 loads various audio files, including natural sounds (such as ocean waves and rainforest sounds), weather sounds (such as light rain and wind sounds), mechanical sounds (such as fan sounds and air conditioner sounds), and artificially synthesized sounds (such as broadband white noise and pink noise). The system randomizes the playback order, with each type played for 5 seconds and interspersed with a 10-second silence as a complete trial, and marks the start of playback in the EEG data stream using a trigger signal. During playback, the EEG acquisition module 2 records signals from all channels and writes millisecond-level synchronized timestamps to ensure precise correspondence between data and stimulus events.

[0156] Before proceeding to feature analysis, data processing module 3 first performs a 1-40Hz bandpass filter on the raw EEG data to remove low-frequency drift and high-frequency electromyographic noise. Secondly, independent component analysis (ICA) is used to separate and remove common artifacts such as eye movement and electrocardiogram (ECG) signals, avoiding interference with the calculation of event-related potentials (ERPs) from these non-brain-derived signals. In this embodiment, artifact removal follows the order of "global interference first, local interference second, and target noise last," meaning that power supply interference and power frequency noise affecting the entire frequency band are filtered out first, then localized EMG and eye movement noise is processed, and finally, specific interference signals under target white noise conditions are processed to ensure that the correlation between channels and the overall temporal relationship are not destroyed during artifact removal. In actual computation, to avoid introducing new noise, the weight matrix in the ICA removal process is smoothed, and the removed components are restored to the signal space through inverse transformation, ensuring that the spectral characteristics of the remaining signal are consistent with the original signal.

[0157] After artifact removal, the system extracts a time-domain segment from -500 ms to 500 ms for each trial based on the trigger marker and performs baseline correction, using the average value from -500 ms to 0 ms before stimulation as a reference point to bring the ERP waveform back to a uniform baseline. During the standardization phase, data processing module 3 performs Z-score conversion on the ERP segment of each channel to eliminate initial amplitude differences among different subjects. Subsequently, the system calculates the absolute mean of the ERP amplitude for each trial within a time window of 100 to 500 ms, which serves as the core indicator for subsequent calculation of the Sound Sensitivity Index (NSI). The formula for calculating NSI is: ,in, This represents the average ERP index for the currently evaluated white noise type. σ represents the mean of other types, and σ is the corresponding standard deviation. The NSI is averaged over multiple rounds for each sound type. The calculated NSI value comprehensively reflects the average intensity and stability of the neural response under that sound condition. Next, the ATI is calculated. ,in This represents the mean of the attention index under a specific target white noise condition. σ represents the mean of the index under other conditions, and σ represents the corresponding standard deviation.

[0158] The NSI and ATI indicators are dynamically weighted based on the signal-to-noise ratio (SNR) during the recording process, with the weight determined by the SNR, thus determining the Fusion Score (FS) under different noise conditions. The FS values ​​for all conditions are then subjected to a one-way ANOVA for significance testing, supplemented by post-hoc comparisons using Tukey HSD, to determine whether the difference between the target condition and other conditions reaches statistical significance (p<0.05), ensuring that the screening results are based on scientific evidence rather than random fluctuations.

[0159] After determining the optimal white noise type, the system enters the volume optimization phase. In this embodiment, a volume gradient of 40 to 70 dB is set, with 5 dB intervals between each gradient. The playback-acquisition-processing-analysis process is repeated, and finally, the optimal volume level is determined by combining the ATI and NSI results. Throughout this process, the system continuously adjusts the volume through the playback control module 1 and generates a new trigger marker at the start of each playback, writing it into the EEG data to ensure the comparability of data under different volume conditions.

[0160] In one embodiment, as shown in Figure 8, a target white noise adaptive recommendation method is provided. This embodiment illustrates the method by applying it to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0161] S802: Extract the sound features of the new environment when entering a new environment.

[0162] S804: Obtain historical preference white noise based on the current task scenario. The historical preference white noise is determined based on the white noise adaptive recommendation method based on EEG signal feedback in any of the above embodiments.

[0163] S806: Based on historical preferred white noise and the sound characteristics of the new environment, the target sound characteristics of the new environment are obtained.

[0164] S808: Determines the target white noise corresponding to the new environment based on the target sound characteristics.

[0165] The entire process forms a data closed loop: sound playback triggers an EEG acquisition event, the raw data undergoes preprocessing, feature extraction, and index calculation before entering the decision and optimization module, and the decision and optimization module feeds back the optimized playback parameters to the playback control module based on the statistical results to achieve adaptive adjustment.

[0166] When a user switches from environment A to environment B, the system first collects at least 5 seconds of background sound signal from the new environment B through the ambient microphone, and then performs standardized preprocessing (DC offset removal, bandpass filtering 20Hz–20kHz, and amplitude normalization) to obtain analyzable waveform data.

[0167] Optionally, extracting sound features of the new environment includes: extracting the sound signal of the new environment; preprocessing the sound signal, and extracting features from the preprocessed sound signal to obtain sound features, which include at least one of environmental entropy value, rhythm intensity, and dominant frequency. For example, when switching to a new environment B, the system calculates the feature vector of environment B:

[0168]

[0169] Specifically, based on the historical preferred white noise and the sound characteristics of the new environment, the target sound characteristics of the new environment are obtained, including: obtaining the sound characteristics of the old environment corresponding to the historical preferred white noise; obtaining the difference sound characteristics based on the sound characteristics of the old environment and the historical preferred white noise; and obtaining the target sound characteristics of the new environment based on the difference sound characteristics and the sound characteristics of the new environment.

[0170] Determining the target white noise corresponding to a new environment based on the target sound features includes: determining the pre-calculated sound features of each white noise in the white noise library; calculating the Euclidean weighted distance between the pre-calculated sound features of each white noise in the white noise library and the sound features of the new environment; selecting a number of white noises whose Euclidean weighted distance meets the requirements as candidate white noises; playing each candidate white noise in sequence, and determining the target white noise based on interactive instructions.

[0171] The system maintains an independent set of user preference parameters, including entropy offset, for each typical scenario (resting EEG, task-oriented EEG, and motion EEG). Rhythm offset and main frequency offset F, which are defined as the differences between the target sound features selected by the user in the current scene last time and the environmental features of the scene (e.g., = - ).

[0172] When switching to a new environment B, the system will use the feature vector of environment B. Historical preference vectors for corresponding scenarios By adding element by element, we obtain the target sound feature vector:

[0173]

[0174] in: The periodic score (rhythm intensity) is calculated after extracting the energy envelope from the short-time Fourier transform (STFT). The frequency of the main peak of the energy spectral density.

[0175] get The system then calculates the Euclidean distance between the feature vectors of each candidate white noise in the local sound library and selects the one with the smallest distance as the recommended result. If no perfect match is found, the system returns the three candidates with the smallest distance for the user to listen to and manually fine-tune. The system updates immediately after the user clicks to confirm or adjusts any feature value. The corresponding offset is stored in the configuration file and persistently saved to the user configuration file corresponding to the scene for quick adaptation next time.

[0176] In some optional embodiments, the method further includes: receiving artifact ratio and signal-to-noise ratio data output by the EEG device, and determining data quality based on the artifact ratio and signal-to-noise ratio data; and, if the correlation between the change in the sound features of the new environment and the data quality is greater than a correlation threshold, adjusting the sound features of the new environment to obtain a new target white noise, and storing the new target white noise as a new historical preferred white noise.

[0177] To ensure real-time performance, the background sound detection and feature calculation module operates at a frequency of 1 Hz. If a change in the ambient RMS sound pressure level exceeds 6 dB SPL, or a change in entropy exceeds 0.2, the feature vector recalculation and recommendation process is triggered. In wearable EEG acquisition, the system receives signal quality metrics (such as artifact ratio and SNR) from the device in real-time via Bluetooth. When a data quality degradation is detected that is correlated with changes in environmental characteristics (Pearson correlation coefficient > 0.5), the system automatically adjusts the H, R, and F parameters to optimize EEG data stability. For example, in task-oriented EEG, when external rhythmic interference is detected, the system will reduce... To reduce phase interference; in resting-state EEG, once the background high-frequency noise increases, the system will down-adjust. To enhance the shielding effect on low-frequency components; in exercise EEG, if an increase in the subject's heart rate or movement rhythm is detected, the system will increase... Synchronized with the rhythm.

[0178] This embodiment achieves recommendation without relying on complex deep models, requiring only one environmental feature extraction and one sound library distance matching. It can also be equipped with a dynamic weight matrix based on weighted distance to adjust the matching priority of H, R, and F according to the importance of the scene. All offsets and weight configurations of the system can be read from files or databases, which facilitates rapid deployment and reproduction in the laboratory.

[0179] This application effectively eliminates non-brain-derived noise such as eye movement and electromyography, while accurately extracting stable ERP responses and EEG power ratios under different white noise conditions, significantly improving the accuracy and reliability of the analysis. Implementation results show that this system can quickly determine the optimal white noise type and volume for an individual in laboratory, office, and nighttime sleep environments, and the scheme has high repeatability, capable of being independently reproduced entirely based on the above description.

[0180] This application utilizes the neurophysiological responses of subjects under different white noise conditions as objective evidence. Through the collaborative operation of modular hardware and highly automated software, it achieves personalized adaptation and dynamic optimization of white noise type and playback volume. This system addresses individual differences in sound sensitivity, cognitive state, emotional regulation, and sleep response by integrating sound stimulus presentation, real-time EEG acquisition, signal feature extraction, index calculation, and result feedback into a closed-loop framework. This allows for the provision of a sound environment solution that best matches the individual's neural response characteristics in a short time. The entire system consists of a white noise playback control module, an EEG acquisition module, a data processing module, a decision-making and optimization module, and a user interface module. These modules communicate in real-time via time synchronization signals and a high-speed data bus, ensuring alignment of sound events with EEG recordings within milliseconds and maximizing the reliability of the analysis results.

[0181] Each white noise sound is mapped to a three-dimensional feature vector (H, R, F), comprehensively describing the relationship between sound characteristics and usage scenarios. The sound entropy value H reflects complexity, the rhythm intensity R affects attention concentration, and the dominant frequency distribution F determines the sound tone. These features together constitute a refined sound evaluation system, laying the foundation for accurate recommendations. Background sound information is simultaneously collected via microphone during EEG acquisition, enabling rapid matching after scene switching.

[0182] When a user enters or switches to a new environment, the system activates the ambient microphone to collect at least 5 seconds of background audio signal at a sampling rate of 44.1kHz and in 16-bit mono mode. Preprocessing is performed through DC offset removal, second-order Butterworth bandpass filtering (20Hz–20kHz), and amplitude normalization. Then, three-dimensional feature extraction is performed on the signal: First, the audio is segmented into 20ms frames (frame shifted by 10ms), the power spectrum of each frame is calculated and normalized to a probability distribution, and the complexity is calculated using the Shannon entropy formula, with the average of the entire segment yielding the environmental entropy value H_env. Second, the envelope signal is extracted, and the autocorrelation method is used to find the maximum normalized peak value within the 0.5Hz–4Hz rhythm frequency range as the rhythm intensity R_env. Finally, a 4096-point FFT (zero-filled to 8192 points) is performed on the entire signal after adding a Hanning window, and the frequency corresponding to the energy peak is taken as the dominant frequency F_env. The system reads the historical preference offset ΔP=[ΔH,ΔR,ΔF] for the current task scenario (resting, tasking, or moving), adds it to the current environment feature vector E_env=[H_env,R_env,F_env] to obtain the target sound feature P_target=[H_t,R_t,F_t]. Then, it reads the feature vectors of all candidate white noise from the sound library and calculates the weighted Euclidean distance using the scene weight matrix [w_H,w_R,w_F]. The system selects the best match (one or the top three) in ascending order of distance for recommendation and preview. After the user confirms or adjusts the parameters, the system immediately calculates and saves the new offset ΔP_new=P_selected−E_env to the configuration file for that scenario, achieving personalized updates. During operation, the system monitors environmental feature changes at a frequency of 1Hz. If the RMS sound pressure level changes by ≥6dB SPL, the entropy changes by ≥0.2, or the rhythm intensity changes by ≥0.1, recalculation and matching are triggered. In EEG acquisition scenarios, the system also receives the artifact ratio and signal-to-noise ratio output by the device simultaneously. Based on the correlation between environmental feature changes and data quality (>0.5), the closed-loop parameters are fine-tuned. For example, in the resting state, H_t is reduced and the proportion of low frequencies is increased; in the task state, R_t is reduced to maintain medium entropy; and in the motion state, R_t is increased to synchronize with the subject's action rhythm. Finally, the recommended white noise file is sent to the playback module and the current environmental features, target features, matching results, and user operation records are archived for subsequent optimization.

[0183] When switching scenes, in the new environment, the target entropy value of the new sound (new environment entropy value + ΔH) is quickly calculated using a fixed offset (ΔH) of "user preferred sound entropy value - original environment entropy value". Seamless migration is achieved through direct matching or fine-tuning. Alternatively, a more complex approach uses a three-dimensional feature vector (H, R, F) to comprehensively describe the relationship between sound characteristics and the usage scenario. The sound entropy value H reflects complexity, rhythm intensity R affects attention concentration, and the dominant frequency distribution F determines the sound tone. These features together constitute a refined sound evaluation system, laying the foundation for accurate recommendations.

[0184] All operations are centrally scheduled by the control terminal. In the first phase, the user interface will randomly play different types of white noise multiple times, repeating this process at least 40 times. In the second phase, the optimal white noise from the first phase will be randomly played multiple times at different decibel levels. In laboratory mode, this invention utilizes a full-function EEG system and a soundproof room to achieve the highest accuracy assessment, suitable for cognitive neuroscience research and clinical intervention verification. In office mode, portable wireless EEG devices and desktop speakers can be used to quickly determine and recommend white noise types and volumes that improve concentration. In sleep mode, a head-mounted EEG monitoring device and a low-latency audio system are combined to achieve real-time dynamic adjustment of the nighttime environment to optimize sleep onset speed and sleep quality. Compared to existing white noise solutions that rely on subjective questionnaires or uniform playback, this invention, based on real-time neural signal analysis and individualized adaptive decision-making, can provide a quantitatively repeatable optimal solution while ensuring hearing safety (sound level not exceeding 70dB). Furthermore, its modular design enables compatible deployment on different platforms such as wired, wireless, LAN, and cloud, exhibiting high versatility and scalability.

[0185] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0186] Based on the same inventive concept, this application also provides a device for implementing the white noise adaptive recommendation method based on EEG signal feedback as described above. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations of one or more embodiments of the white noise adaptive recommendation device based on EEG signal feedback provided below can be found in the limitations of the white noise adaptive recommendation method based on EEG signal feedback described above, and will not be repeated here.

[0187] In an exemplary embodiment, as shown in FIG9, a white noise adaptive recommendation device based on EEG signal feedback is provided, comprising: a type white noise playback module 901, a type white noise determination module 902, a volume white noise playback module 903, a volume white noise determination module 904, and a preference determination module 905, wherein:

[0188] The white noise playback module 901 is used to control the audio playback device to play different types of white noise, and the volume levels of the different types of white noise are the same.

[0189] The white noise type determination module 902 is used to collect various types of EEG signals under stimulation by different types of white noise, and to determine at least one target type of white noise based on the various types of EEG signals.

[0190] The volume white noise playback module 903 is used to control the audio playback device to play white noise at different volume levels for each target type of white noise;

[0191] The volume white noise determination module 904 is used to collect volume EEG signals under white noise at various volume levels, and determine the white noise at the target volume level based on the volume EEG signals.

[0192] Preference determination module 905 is used to determine white noise of a target volume level under a target type as the starting historical preference white noise based on white noise of at least one target volume level and white noise of at least one target type.

[0193] In one optional embodiment, the white noise playback module 901 and the volume white noise playback module 903 described above are specifically used to control the audio playback device to send identification information to the EEG acquisition device. The identification information carries the sound type and time identifier and is used to identify the EEG signal acquired by the EEG acquisition device when playing each white noise.

[0194] In one optional embodiment, the aforementioned white noise type determination module 902 and volume white noise determination module 904 are specifically used to extract features from the EEG signal for each current white noise to obtain EEG signal features. Specifically, when the target type of white noise is determined, the current white noise is of type , and the EEG signal includes type EEG signals; when the target volume level of white noise is determined, the current white noise is white noise at each volume level, and the EEG signal is volume EEG signals; the white noise evaluation index value is determined based on the EEG signal features; the dynamic weight of each white noise evaluation index value is determined based on the white noise evaluation index value; the evaluation result of the current white noise is obtained based on the dynamic weight and the white noise evaluation index value; and the white noise to be screened is determined based on the evaluation results of each current white noise. When the target type of white noise is determined, the white noise to be screened is white noise of at least one target type; when the target volume level of white noise is determined, the white noise to be screened is white noise at at least one target volume level.

[0195] In one optional embodiment, the aforementioned white noise determination module 902 and volume white noise determination module 904 are specifically configured to, for each current white noise, obtain, based on EEG signal characteristics, the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise; obtain an initial sound sensitivity index value based on the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise; normalize the initial sound sensitivity index value to obtain a target sound sensitivity index value; for each current white noise, determine the target pre-power ratio of different waves based on EEG signal characteristics, and determine an initial attention index value based on the target pre-power ratio of different waves; normalize the initial attention index value of the current white noise to obtain a normalized attention index value; obtain candidate attention index values ​​based on the normalized attention index values ​​of each current white noise, and normalize the candidate attention index values ​​to obtain a target attention index value.

[0196] In one optional embodiment, the white noise determination module 902 and the volume white noise determination module 904 described above are specifically used to obtain the signal-to-noise ratio of each indicator based on the white noise evaluation index value; and to determine the dynamic weight of each white noise evaluation index based on the signal-to-noise ratio of each indicator.

[0197] In an exemplary embodiment, as shown in FIG10, a target white noise adaptive recommendation device is provided, including: an extraction module 1001, a historical preference acquisition module 1002, a sound feature determination module 1003, and a target white noise determination module 1004, wherein:

[0198] Extraction module 1001 is used to extract the sound features of a new environment when entering a new environment;

[0199] The historical preference acquisition module 1002 is used to acquire historical preference white noise based on the current task scenario. The historical preference white noise is determined based on the white noise adaptive recommendation device based on EEG signal feedback in any of the above embodiments.

[0200] The sound feature determination module 1003 is used to obtain the target sound features of the new environment based on historical preferred white noise and the sound features of the new environment;

[0201] The target white noise determination module 1004 is used to determine the target white noise corresponding to a new environment based on the target sound characteristics.

[0202] In one optional embodiment, the extraction module 1001 is specifically used to extract sound signals from the new environment; preprocess the sound signals, and extract features from the preprocessed sound signals to obtain sound features, the sound features including at least one of environmental entropy value, rhythm intensity, and dominant frequency.

[0203] The aforementioned sound feature determination module 1003 is specifically used to obtain the sound features of the old environment corresponding to the historical preferred white noise; based on the sound features of the old environment and the historical preferred white noise, obtain the difference sound features; based on the difference sound features and the sound features of the new environment, obtain the target sound features of the new environment.

[0204] The aforementioned target white noise determination module 1004 is specifically used to determine the pre-calculated sound features of each white noise in the white noise library; calculate the Euclidean weighted distance between the pre-calculated sound features of each white noise in the white noise library and the sound features of the new environment; select a number of white noises whose Euclidean weighted distance meets the requirements as candidate white noises; play each candidate white noise in sequence, and determine the target white noise based on the interactive instructions.

[0205] In some optional embodiments, the above-described apparatus further includes: an adjustment module for receiving artifact ratio and signal-to-noise ratio data output by the EEG device, and determining data quality based on the artifact ratio and signal-to-noise ratio data; when the correlation between the change in the sound features of the new environment and the data quality is greater than a correlation threshold, adjusting the sound features of the new environment to obtain a new target white noise, and storing the new target white noise as a new historical preferred white noise.

[0206] The modules in the aforementioned white noise adaptive recommendation device based on EEG signal feedback can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0207] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram is shown in Figure 11. The computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a white noise adaptive recommendation method based on EEG signal feedback. The display unit of the computer device is used to form a visually visible image and may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0208] Those skilled in the art will understand that the structure shown in FIG11 is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. A specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments. In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program, when executed by a processor, implements the steps in the above-described method embodiments. In one embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0209] It should be fully understood that the user information involved in this application (including but not limited to user physiological information, user personal information, etc.) is information and data authorized by the user or fully authorized by all parties. The use of user information shall comply with privacy policies and practices that are often considered to meet or exceed industry or government requirements for maintaining user privacy. The collection, use and processing of related data shall comply with relevant laws, regulations and standards, and provide corresponding operation access points for users to choose to authorize or refuse.

[0210] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one of relational and non-relational databases. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units, artificial intelligence (AI) processors, etc., and are not limited to these. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this application. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. A white noise adaptive recommendation method based on EEG signal feedback, characterized in that, The method further includes: controlling an audio playback device to play different types of white noise, wherein the volume levels of the different types of white noise are the same; acquiring various types of electroencephalogram (EEG) signals under the stimulation of the different types of white noise, and determining at least one target type of white noise based on the EEG signals of each type; controlling the audio playback device to play white noise at different volume levels for each target type of white noise; acquiring volume EEG signals under the white noise at each volume level, and determining white noise at a target volume level based on the volume EEG signals; and determining a target volume level of white noise under the target type as a starting point based on the determined white noise at at least one target volume level and white noise of at least one target type. Historical preference for white noise; the target type of white noise and the white noise at the target volume level are determined by the following methods: for each current white noise, feature extraction is performed on the EEG signal to obtain EEG signal features, wherein when the target type of white noise is determined, the current white noise is white noise of that type, and the EEG signal includes the type of EEG signal; when the target volume level of white noise is determined, the current white noise is the white noise at each of the volume levels, and the EEG signal is the volume EEG signal; the white noise evaluation index value is determined based on the EEG signal features; the dynamic weight of each white noise evaluation index is determined based on the white noise evaluation index value; and the dynamic weight and Each of the aforementioned white noise evaluation index values ​​yields the evaluation result of the current white noise; based on the evaluation results of each of the current white noises, white noise to be screened is determined. If the target type of white noise is determined, the white noise to be screened is white noise of at least one target type; if the target volume level of white noise is determined, the white noise to be screened is white noise of at least one target volume level; determining the white noise evaluation index value based on the EEG signal characteristics includes: for each current white noise, based on the EEG signal characteristics, obtaining the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise; based on the current white noise... The amplitude variance of the corresponding EEG signal, the amplitude variance of the EEG signal corresponding to the non-current white noise, and the average value of the EEG signal corresponding to the non-current white noise are used to obtain an initial sound sensitivity index value. The initial sound sensitivity index value is then normalized to obtain a target sound sensitivity index value. For each current white noise, a target pre-power ratio for different waves is determined based on the characteristics of the EEG signal, and an initial attention index value is determined based on the target pre-power ratio for different waves. The initial attention index value for the current white noise is then normalized to obtain a normalized attention index value. Candidate attention index values ​​are obtained based on the normalized attention index values ​​for each current white noise, and the candidate attention index values ​​are then normalized to obtain a target attention index value.

2. The method according to claim 1, characterized in that, The method further includes: controlling the audio playback device to send identification information to the EEG acquisition device, the identification information carrying a sound type and time identifier, the identification information being used to identify the EEG signals acquired by the EEG acquisition device when playing various white noises.

3. The method according to claim 1, characterized in that, The step of determining the dynamic weight of each white noise evaluation index based on the value of each white noise evaluation index includes: obtaining the signal-to-noise ratio of each index based on the value of each white noise evaluation index; and determining the dynamic weight of each white noise evaluation index based on the signal-to-noise ratio of each index.

4. A target white noise adaptive recommendation method based on the white noise adaptive recommendation method based on EEG signal feedback as described in any one of claims 1 to 3, characterized in that, The method includes: extracting sound features of the new environment when entering a new environment; obtaining historical preferred white noise based on the current task scenario; obtaining target sound features of the new environment based on the historical preferred white noise and the sound features of the new environment; and determining target white noise corresponding to the new environment based on the target sound features.

5. The method according to claim 4, characterized in that, The step of extracting the sound features of the new environment includes: extracting the sound signal of the new environment; preprocessing the sound signal and extracting features from the preprocessed sound signal to obtain sound features, wherein the sound features include at least one of environmental entropy value, rhythm intensity, and dominant frequency; the step of obtaining the target sound features of the new environment based on the historical preferred white noise and the sound features of the new environment includes: obtaining the sound features of the old environment corresponding to the historical preferred white noise; obtaining the difference sound features based on the sound features of the old environment and the historical preferred white noise; obtaining the target sound features of the new environment based on the difference sound features and the sound features of the new environment; the step of determining the target white noise corresponding to the new environment based on the target sound features includes: determining the pre-calculated sound features of each white noise in the white noise library; calculating the Euclidean weighted distance between the pre-calculated sound features of each white noise in the white noise library and the sound features of the new environment; selecting a target number of white noises whose Euclidean weighted distance meets the requirements as candidate white noises; playing each candidate white noise in sequence, and determining the target white noise based on interactive instructions.

6. The method according to claim 5, characterized in that, The method further includes: receiving artifact ratio and signal-to-noise ratio data output by the EEG device, and determining data quality based on the artifact ratio and signal-to-noise ratio data; when the correlation between the change in the sound features of the new environment and the data quality is greater than a correlation threshold, adjusting the sound features of the new environment to obtain a new target white noise, and storing the new target white noise as a new historical preferred white noise.

7. A white noise adaptive recommendation device based on EEG signal feedback, characterized in that, The device includes: a white noise type playback module for controlling an audio playback device to play different types of white noise, wherein the volume levels of the different types of white noise are the same; a white noise type determination module for acquiring various types of EEG signals under the stimulation of the different types of white noise, and determining at least one target type of white noise based on the EEG signals of each type; a white noise volume playback module for controlling the audio playback device to play white noise at different volume levels for each target type of white noise; a white noise volume determination module for acquiring volume EEG signals under the white noise at each volume level, and determining the white noise at a target volume level based on the volume EEG signals; and a preference determination module. For determining white noise of a target volume level under a target type as the starting historical preferred white noise, based on at least one target volume level of white noise and at least one target type of white noise; the type white noise determination module and the volume white noise determination module are specifically used to perform feature extraction on the EEG signal for each current white noise to obtain EEG signal features, wherein when the target type of white noise is determined, the current white noise is white noise of that type, and the EEG signal includes the type EEG signal; when the target volume level of white noise is determined, the current white noise is the white noise at each of the volume levels, and the EEG signal is the volume EEG signal; based on the brain The evaluation criteria for each white noise index are determined based on the characteristics of the electrical signal. A dynamic weight for each white noise evaluation index is determined based on its value. An evaluation result for the current white noise is obtained based on the dynamic weight and the evaluation values ​​of each white noise index. White noise to be screened is determined based on the evaluation results of the current white noise. If the target type of white noise is determined, the white noise to be screened is white noise of at least one target type. If the target volume level of white noise is determined, the white noise to be screened is white noise of at least one target volume level. Specifically, the white noise type determination module and the white noise volume determination module are used to, for each current white noise, obtain the corresponding white noise based on the characteristics of the EEG signal. The amplitude variance of the EEG signal, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise are calculated. Based on the amplitude variance of the EEG signal corresponding to the current white noise, the amplitude variance of the EEG signal not corresponding to the current white noise, and the average value of the EEG signal not corresponding to the current white noise, an initial sound sensitivity index value is obtained. The initial sound sensitivity index value is normalized to obtain a target sound sensitivity index value. For each current white noise, a target pre-power ratio for different waves is determined based on the characteristics of the EEG signal, and an initial attention index value is determined based on the target pre-power ratio for different waves. The initial attention index value for the current white noise is normalized to obtain a normalized attention index value.Candidate attention index values ​​are obtained based on the normalized attention index values ​​of each current white noise, and the candidate attention index values ​​are then normalized to obtain the target attention index value.

8. The apparatus according to claim 7, characterized in that, The white noise type determination module and the white noise volume determination module are specifically used to control the audio playback device to send identification information to the EEG acquisition device. The identification information carries a sound type and a time identifier. The identification information is used to identify the EEG signals acquired by the EEG acquisition device when playing each type of white noise.

9. The apparatus according to claim 7, characterized in that, The white noise type determination module and the white noise volume determination module are specifically used to obtain the signal-to-noise ratio of each indicator based on the white noise evaluation index value; and to determine the dynamic weight of each white noise evaluation index based on the signal-to-noise ratio of each indicator.

10. A target white noise adaptive recommendation device based on the white noise adaptive recommendation device based on EEG signal feedback as described in any one of claims 7 to 9, characterized in that, The device includes: an extraction module for extracting sound features of a new environment upon entering a new environment; a history preference acquisition module for acquiring historical preference white noise based on the current task scenario; a sound feature determination module for obtaining target sound features of the new environment based on the historical preference white noise and the sound features of the new environment; and a target white noise determination module for determining the target white noise corresponding to the new environment based on the target sound features.

Images