Audio content pushing method and system

Abstract

The embodiment of the invention discloses an audio content pushing method and system. The method comprises the steps that electroencephalogram signals of a user are collected; recognizing the health state of the user based on the electroencephalogram signal; according to the health state, matching with the audio content in the audio content set, and calculating a first matching degree of each audio content and the health state; and pushing the audio content to the user according to the first matching degree, and repeating the steps until the health state reaches a preset value. According to the technical scheme provided by the invention, the personalized audio content can be pushed according to the health states of different users, and the technical problem that the audio is difficult to adaptively push to each user in the prior art is relieved; in addition, the audio content does not need to be manually researched and tested, and the technical problems that an existing music healing system is long in development period and high in cost are solved.

Description

Technical Field

[0001] The present application relates to the technical field of artificial intelligence systems, and specifically to an audio content push method and system. Background Art

[0002] In the prior art, music therapy devices based on brain-computer interfaces collect users' EEG data and perform music effect tests to form treatment plans that promote users' physical and mental health and improve their mood. These plans require professional music research and effect testing, as well as the cooperation of users to achieve a certain healing effect. However, music research and development requires the investment of professionals and a lot of time, and some music cannot be used freely due to copyright restrictions. Therefore, the content of the audio resource library formed is limited, and it is difficult to cover a wider range of audio content needs. In addition, due to the limited number of samples of test users, it is difficult for existing music therapy systems to adaptively push audio to each user.

[0003] In view of this, there is an urgent need to provide users with a system and method for pushing personalized audio content to alleviate existing technical problems. Summary of the invention

[0004] In view of this, the purpose of the embodiments of the present application is to provide an audio content push method and system, which, based on the real-time data of the brain-computer interface device, uses an intelligent recommendation algorithm to generate personalized audio content push for each user, adapting to the needs and preferences of different individuals, so as to alleviate existing technical problems.

[0005] According to some embodiments, the first aspect of the present application provides an audio content push method, including: collecting the user's electroencephalogram (EEG) signal; identifying the user's health status based on the EEG signal; matching the audio content in the audio content set according to the health status, and calculating the first matching degree between each audio content and the health status; pushing the audio content to the user according to the first matching degree; repeating the above steps until the health status reaches a preset value.

[0006] In one embodiment, the step of collecting the user's EEG signal includes: collecting at least one EEG signal of the user's brain among alpha waves, beta waves, gamma waves, theta waves and delta waves; and calculating at least one EEG parameter related to concentration, stress value, emotion value, fatigue, relaxation and mindfulness based on at least one EEG signal.

[0007] In one embodiment, the step of identifying the user's health status includes: normalizing EEG signals and / or EEG parameters; establishing a health status prediction model based on the EEG signals and / or EEG parameters using a linear regression algorithm, wherein the prediction model includes a penalty term for variable screening and complexity adjustment; the health status prediction model outputs the health status.

[0008] In one embodiment, the steps of identifying the health status of the user include: feature extraction, extracting the first feature of the EEG signal in the time domain and / or frequency domain; feature conversion, sending the first feature into a deep learning neural network and outputting a second feature, the deep learning neural network includes a fully connected layer and a product layer; feature dimensionality reduction, using principal component analysis to reduce the dimension of the second feature to generate a third feature; health status output, based on the third feature, using a support vector machine to predict the health status.

[0009] In one embodiment, the method further includes: obtaining user behavior data on the audio content set; calculating a second matching degree between each audio content in the audio content set and the user based on the behavior data; and pushing the audio content to the user based on the first matching degree and / or the second matching degree.

[0010] In one embodiment, the step of calculating the second matching degree between each audio content in the audio content set and the user includes: performing cluster analysis based on behavioral data and basic information of the user to construct a digital portrait of the user; and calculating the second matching degree between each audio content in the audio content set and the user based on the digital portrait.

[0011] In one embodiment, cluster analysis is performed based on behavioral data and basic information of users, and the steps of constructing a digital portrait of the user include: based on the basic information of the users, statistical analysis is performed on different users and the distribution of user behavioral data to form a data sequence; hierarchical cluster analysis is performed on the data sequence, a distance matrix is ​​constructed using Euclidean distance, and the distance between data in the data sequence is calculated; two data with the closest distance are merged to form a cluster, and the distance matrix is ​​updated to determine the distance between the new cluster and other data; clusters are repeatedly merged and the distance matrix is ​​updated until the number of clusters for all data points reaches a preset value; and a digital portrait is constructed according to the number of clusters.

[0012] In one embodiment, the step of calculating the second matching degree between each audio content in the audio content set and the user also includes: performing feature extraction on the audio content in the audio content set to obtain a fourth feature, and performing feature extraction on the behavior data to obtain a fifth feature; performing correlation analysis on the fourth feature and the fifth feature using a logistic regression model, and calculating the second matching degree between each audio content in the audio content set and the user.

[0013] In one embodiment, the step of extracting features from the audio content of the audio content set to obtain the fourth feature includes: extracting the Mel spectrum of the audio content based on the melody, rhythm, harmony, and emotion of the audio content to form the fourth feature.

[0014] In one embodiment, the step of extracting features from the behavior data to obtain the fifth feature includes: based on the user's access history, rating and collection behavior of the audio content, data splicing and weighting are performed to form the fifth feature.

[0015] In one embodiment, it also includes: dynamically generating custom audio based on health status and / or behavior data, the custom audio includes generating symbolic music through text description, generating different emotional music, generating lyrics and melodies of different musical forms; and pushing the custom audio to users.

[0016] According to some embodiments, the second aspect of the present application provides an audio content push system, characterized in that it includes: an EEG device module, which is used to collect the user's EEG signals; a data analysis module, which is used to identify the user's health status based on the EEG signals; according to the health status, matching with the audio content in the audio content set, and calculating the first matching degree between each audio content and the health status; a treatment plan module, which is used to push audio content to the user according to the first matching degree; according to the health status, changing the pushed audio content until the health status reaches a preset value.

[0017] In one embodiment, it also includes: a music generation module, which is used to dynamically generate custom audio according to health status and / or behavior data, the custom audio includes generating symbolic music through text description, generating different emotional music, generating lyrics and melodies of different musical forms; and pushing custom audio to users.

[0018] In one embodiment, the EEG device module is further configured to: collect at least one EEG signal of alpha wave, beta wave, gamma wave, theta wave and delta wave from the user's brain; calculate at least one EEG parameter related to concentration, stress value, emotion value, fatigue, relaxation and mindfulness based on at least one EEG signal.

[0019] In one embodiment, the treatment plan module is further configured to: obtain the user's behavioral data on the audio content set; based on the behavioral data, calculate the second matching degree between each audio content in the audio content set and the user; based on the first matching degree and / or the second matching degree, push the audio content to the user.

[0020] The disclosed embodiments may have / at least have the following advantages: by identifying the user's health status based on EEG signals, and then matching the health status with the audio content in the audio content set, personalized audio content can be pushed according to the health status of different users, without the need for manual research and testing of the audio content, and also alleviating the technical problems of long development cycle and high cost of existing music therapy systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flowchart of an audio content push method according to an embodiment of the present application;

[0022] Figure 2 It is a logical structure diagram of an audio content push system according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below in conjunction with specific implementations and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present application. In addition, in the following description, the description of known structures and technologies is omitted to avoid unnecessary confusion of the concepts of the present application.

[0024] The schematic diagram of the layer structure according to the embodiment of the present application is shown in the accompanying drawings. These figures are not necessarily drawn to scale, and some details are magnified and some details may be omitted for the purpose of clarity. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may further design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0025] Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. The present application will be described in more detail below with reference to the accompanying drawings. In each of the drawings, the same elements are represented by similar reference numerals. For clarity, the various parts in the drawings are not drawn to scale.

[0026] Figure 1 It is a flowchart of an audio content pushing method according to an embodiment of the present application.

[0027] refer to Figure 1 , an audio content pushing method provided in this embodiment includes the following four steps.

[0028] Step S1: Collecting the user's EEG signal. It should be noted that collecting EEG signals is a method of recording brain activity using electrophysiological indicators. When the user's brain is active, the postsynaptic potentials of a large number of neurons that occur synchronously are integrated to form an EEG signal, which is the overall reflection of the electrophysiological activity of brain nerve cells on the cerebral cortex or scalp surface, and can be collected using an EEG recorder.

[0029] In one embodiment, the step of collecting the user's EEG signal includes: collecting at least one EEG signal of the user's brain among alpha waves, beta waves, gamma waves, theta waves and delta waves; and calculating at least one EEG parameter related to concentration, stress value, emotion value, fatigue, relaxation and mindfulness based on at least one EEG signal.

[0030] In one embodiment, concentration can be represented by the power of theta waves divided by the power of beta waves; stress value can be represented by the absolute power or relative power of theta waves; emotion value can be characterized by the root mean square of the wavelet coefficients of the alpha wave band; fatigue can be represented by the power of theta waves divided by the power of alpha waves; relaxation can be represented by the power of alpha waves; and mindfulness can be represented by the relative power of theta waves and delta waves.

[0031] Step S2: Identify the health status of the user. Based on the EEG signal collected in step S1, identify the health status of the user. For example, the health status of the user can be identified through the time domain or frequency domain features of the EEG signal.

[0032] In one embodiment, the step of identifying the user's health status includes: normalizing EEG signals and / or EEG parameters; establishing a health status prediction model based on the EEG signals and / or EEG parameters using a linear regression algorithm, wherein the prediction model includes a penalty term for variable screening and complexity adjustment; the health status prediction model outputs the health status.

[0033] Exemplarily, the normalization of EEG signal features adopts the "maximum-minimum method", which can standardize the EEG signal data to make it more suitable for further analysis.

[0034] For example, the Max-Min formula is as follows:

[0035] Among them, min(x(n)) represents the minimum value of the sample data x(n), max(x(n)) represents the maximum value of the sample data x(n), and y(k) is the normalized value of x(k).

[0036] Exemplarily, the linear regression algorithm can be a LASSO regression model, which is used to analyze the normalized data and establish a health index model. LASSO regression is a reduction method of linear regression, which limits the weight by adding a penalty term to perform variable screening and complexity adjustment. The loss function of the LASSO regression model can be expressed as: Where L(β) is the loss function, β is the model parameter, m is the number of samples, n is the number of features, and y i is the observed value of the i-th sample, X ij is the value of the i-th sample on the j-th feature, β j is the model coefficient corresponding to the jth feature, and λ is the regularization parameter, which controls the strength of the penalty term. The larger the value of λ, the stronger the penalty for the coefficient, which may cause more coefficients to be compressed to 0. Using LASSO regression, the purpose of health status prediction can be achieved through feature extraction and model simplification.

[0037] In one embodiment, the steps of identifying the health status of the user include: feature extraction, extracting the first feature of the EEG signal in the time domain and / or frequency domain; feature conversion, sending the first feature into a deep learning neural network and outputting a second feature, the deep learning neural network includes a fully connected layer and a product layer; feature dimensionality reduction, using principal component analysis to reduce the dimension of the second feature to generate a third feature; health status output, based on the third feature, using a support vector machine to predict the health status.

[0038] Exemplarily, the first feature of the EEG signal can be a time domain feature, such as the statistical features of the data, mean, variance, etc.; the first feature can also be a frequency domain feature, such as the power spectrum density in the frequency domain, the energy value of a specific frequency band; the first feature can be represented in the form of a vector or a tensor. The fully connected layer and the accumulation layer of the deep learning neural network process and transform the first feature to form a second feature to characterize the health state. Principal component analysis is used to map the second feature from a high-dimensional feature to a low-dimensional space to make it easier to identify and interpret the health state.

[0039] It should be noted that both the LASSO regression model and the support vector machine prediction of health status use trained models. In the process of training the model, the cross entropy loss function is used to measure the difference between the model's prediction and the actual health status, and the gradient descent optimization algorithm is used to adjust the model parameters to minimize the loss function.

[0040] It is also important to note that the output of the model can be evaluated and interpreted to ensure that it accurately reflects the health status. For example, the precision and recall evaluation indicators can be calculated, and the prediction results of the model can be explained.

[0041] Step S3: Calculate the first matching degree between each audio content and the health state. For example, the health state can be represented by a relaxation degree, for example, the relaxation degree is 0-25 for high stress state, 25-50 for medium stress, 50-75 for low stress, and 75-100 for no stress state. The audio content in the audio content set can be matched to users with different stress levels by extracting its melody and rhythm.

[0042] In one embodiment, the method further includes: obtaining user behavior data on the audio content set; calculating a second matching degree between each audio content in the audio content set and the user based on the behavior data; and pushing the audio content to the user based on the first matching degree and / or the second matching degree.

[0043] It should be noted that the calculation of the first matching degree is based on the user's EEG signal state, which characterizes the user's real-time state; the calculation of the second matching degree is based on the user's historical behavioral data, which characterizes the user's historical behavior and habits. The first matching degree and the second matching degree can be weighted to make a decision on pushing audio content. In addition, when calculating the second matching degree, the behavioral data based on it can also be weighted. For example, recent historical records may have a greater impact on the current health status, so a higher weight can be given.

[0044] In one embodiment, the step of calculating the second matching degree between each audio content in the audio content set and the user includes: performing cluster analysis based on behavioral data and basic information of the user to construct a digital portrait of the user; and calculating the second matching degree between each audio content in the audio content set and the user based on the digital portrait.

[0045] It should be noted that the digital portrait model can combine users' basic information and behavioral data to cluster users into different groups. Each group has a specific label, which is used to characterize which audio content has a high preference. For example, the label can be "young / passionate music", "older / soothing music", etc.

[0046] In one embodiment, cluster analysis is performed based on behavioral data and basic information of users, and the steps of constructing a digital portrait of the user include: based on the basic information of the users, statistical analysis is performed on different users and the distribution of user behavioral data to form a data sequence; hierarchical cluster analysis is performed on the data sequence, a distance matrix is ​​constructed using Euclidean distance, and the distance between data in the data sequence is calculated; two data with the closest distance are merged to form a cluster, and the distance matrix is ​​updated to determine the distance between the new cluster and other data; clusters are repeatedly merged and the distance matrix is ​​updated until the number of clusters for all data points reaches a preset value; and a digital portrait is constructed according to the number of clusters.

[0047] For example, the basic information of the user may include age characteristics, gender characteristics, place of origin characteristics, etc., and the behavior data may include browsing characteristics, click characteristics, collection characteristics, comment characteristics, forwarding characteristics, etc. of the audio content. The data in the data sequence may be a vector obtained by normalizing or weighting the above characteristics, and the Euclidean distance between the data may be calculated by the Euclidean distance between the vectors.

[0048] In one embodiment, the step of calculating the second matching degree between each audio content in the audio content set and the user also includes: performing feature extraction on the audio content in the audio content set to obtain a fourth feature, and performing feature extraction on the behavior data to obtain a fifth feature; performing correlation analysis on the fourth feature and the fifth feature using a logistic regression model, and calculating the second matching degree between each audio content in the audio content set and the user.

[0049] In one embodiment, the step of extracting features from the audio content of the audio content set to obtain the fourth feature includes: extracting the Mel spectrum of the audio content based on the melody, rhythm, harmony, and emotion of the audio content to form the fourth feature.

[0050] In one embodiment, the step of extracting features from the behavior data to obtain the fifth feature includes: based on the user's access history, rating and collection behavior of the audio content, data splicing and weighting are performed to form the fifth feature.

[0051] Exemplarily, the logistic regression model can be a random logistic regression model and a LightGBM model to analyze the user's preference for audio content. Random logistic regression is one of the methods used for feature screening in the stability selection method; stability selection is a method based on the combination of subsampling and selection algorithm, and the selection algorithm can be regression, SVM or other methods. The main idea is to run the feature selection algorithm on different data subsets and feature subsets, repeat it continuously, and finally summarize the feature selection results. For example, count the frequency of a feature being considered an important feature: the number of times it is selected as an important feature, such as "the number of times the audio content is accessed during a period of time." The lightGBM model is an implementation of a gradient boosting decision tree. Its essential principle is to use decision tree training integration to obtain the optimal model.

[0052] Step S4: Pushing audio content to the user. Exemplarily, the music playing unit and the Bluetooth transmission unit can be integrated in the EEG signal acquisition device. The music playing unit plays music by bone conduction, transmits music data by Bluetooth, and transmits audio data to other Bluetooth devices through the Bluetooth transmission unit.

[0053] After executing step S4, the EEG signal of the user after listening to the audio content can also be collected by the EEG signal collection device to evaluate the user's health status. If the user's health status does not reach the preset value, the above steps need to be iterated until the user's health status reaches the preset value.

[0054] In one embodiment, it also includes: dynamically generating custom audio based on health status and / or behavior data, the custom audio includes generating symbolic music through text description, generating different emotional music, generating lyrics and melodies of different musical forms; and pushing the custom audio to users.

[0055] For example, custom audio can use the DeepRapper model to generate lyrics, use the EmoGen model to generate a variety of emotional music, use the Musecoco model to generate symbolic music through a text description, and use the MeloForm model to generate melodies of different musical forms.

[0056] It should be noted that the DeepRapper model is a rap generation system based on the Transformer model, which can simulate rhyme and rhythm to generate lyrics; the EmoGen model can generate audio content based on emotions. Customized audio can alleviate the technical problem of limited content in existing audio resource libraries and achieve the technical effect of dynamically and flexibly generating audio content.

[0057] Figure 2 It is a logical structure diagram of an audio content push system according to an embodiment of the present application.

[0058] refer to Figure 2 The embodiment of the present application also provides an audio content push system, including: an EEG device module 101, which is used to collect EEG signals of users; a data analysis module 102, which is used to identify the health status of the user based on the EEG signals; according to the health status, match the audio content in the audio content set, and calculate the first matching degree between each audio content and the health status; a treatment plan module 103, which is used to push audio content to the user according to the first matching degree; according to the health status, change the pushed audio content until the health status reaches a preset value.

[0059] In one embodiment, it also includes: a music generation module, which is used to dynamically generate custom audio according to health status and / or behavior data, the custom audio includes generating symbolic music through text description, generating different emotional music, generating lyrics and melodies of different musical forms; and pushing custom audio to users.

[0060] In one embodiment, the EEG device module 101 is further configured to: collect at least one EEG signal of alpha wave, beta wave, gamma wave, theta wave and delta wave from the user's brain; calculate at least one EEG parameter related to concentration, stress value, emotion value, fatigue, relaxation and mindfulness based on at least one EEG signal.

[0061] In one embodiment, the music playing unit and the Bluetooth transmission unit can be integrated in the EEG device module 101. The music playing unit plays music by bone conduction, transmits music data by Bluetooth, and transmits audio data with other Bluetooth devices through the Bluetooth transmission unit.

[0062] In one embodiment, the treatment plan module 103 is further configured to: obtain the user's behavior data on the audio content set; based on the behavior data, calculate the second matching degree between each audio content in the audio content set and the user; based on the first matching degree and / or the second matching degree, push the audio content to the user.

[0063] It should be understood that the above specific embodiments of the present application are only used to illustrate or explain the principles of the present application, and do not constitute a limitation to the present application. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present application should be included in the protection scope of the present application. In addition, the claims attached to the present application are intended to cover all changes and modifications that fall within the scope and boundaries of the attached claims, or the equivalent forms of such scope and boundaries.

Claims

1. A method for pushing audio content, characterized in that: include: Collecting the user's EEG signals; Based on the EEG signal, identifying the health status of the user; According to the health status, matching is performed with audio content in the audio content set, and calculating a first matching degree between each of the audio content and the health status; Pushing the audio content to the user according to the first matching degree; Repeat the above steps until the health status reaches a preset value.

2. The method for pushing audio content according to claim 1, characterized in that: The step of collecting the user's electroencephalogram signal comprises: Collecting at least one of the EEG signals of alpha wave, beta wave, gamma wave, theta wave and delta wave from the user's brain; At least one EEG parameter related to concentration, stress, emotion, fatigue, relaxation and mindfulness is calculated based on at least one of the EEG signals.

3. The method for pushing audio content according to claim 2, characterized in that: The step of identifying the health status of the user comprises: Normalizing the EEG signal and / or EEG parameters; Based on the EEG signal and / or EEG parameters, a health status prediction model is established using a linear regression algorithm, wherein the prediction model includes a penalty term for variable screening and complexity adjustment; the health status prediction model outputs the health status.

4. The method for pushing audio content according to claim 2, characterized in that: The step of identifying the health status of the user comprises: Feature extraction, extracting a first feature of the EEG signal in time domain and / or frequency domain; feature Transformation, sending the first feature into the deep learning neural network, outputting the second feature, The deep learning neural network includes a fully connected layer and a cumulative layer; Feature dimensionality reduction, using principal component analysis to reduce the dimension of the second feature to generate a third feature; The health status is outputted based on the third feature and a support vector machine is used to predict the health status.

5. The method for pushing audio content according to claim 1, characterized in that: Also includes: Acquire user behavior data on the audio content set; Based on the behavior data, calculating a second matching degree between each audio content in the audio content set and the user; Based on the first matching degree and / or the second matching degree, the audio content is pushed to the user.

6. The method for pushing audio content according to claim 5, characterized in that: The step of calculating the second matching degree between each audio content in the audio content set and the user comprises: Performing cluster analysis based on the behavior data and basic information of the user to construct a digital portrait of the user; Based on the digital portrait, a second matching degree between each audio content in the audio content set and the user is calculated.

7. The method for pushing audio content according to claim 6, characterized in that: The steps of performing cluster analysis based on the behavior data and the basic information of the user and constructing a digital portrait of the user include: Based on the basic information of users, statistical analysis is performed on different users and the distribution of user behavior data to form a data sequence; Performing hierarchical clustering analysis on the data sequence, constructing a distance matrix using Euclidean distance, and calculating the distance between data in the data sequence; Merge the two closest data to form a cluster, and update the distance matrix to determine the distance between the new cluster and other data; Repeat merging clusters and updating the distance matrix until the number of clusters for all data points reaches the preset value; The digital portrait is constructed according to the number of clusters.

8. The method for pushing audio content according to claim 5, characterized in that: The step of calculating the second matching degree between each audio content in the audio content set and the user also includes: Extracting features from the audio content in the audio content set to obtain a fourth feature, and extracting features from the behavior data to obtain a fifth feature; A logistic regression model is used to perform correlation analysis on the fourth feature and the fifth feature, and a second matching degree between each audio content in the audio content set and the user is calculated.

9. The method for pushing audio content according to claim 8, characterized in that: The step of extracting features from the audio content of the audio content set to obtain a fourth feature includes: Based on the melody, rhythm, harmony, and emotion of the audio content, the Mel spectrum of the audio content is extracted to form the fourth feature.

10. The method for pushing audio content according to claim 8, characterized in that: The step of extracting features from the behavior data to obtain the fifth feature includes: Based on the user's access history, rating and collection behavior of the audio content, the fifth feature is formed after data splicing and weighting.

11. The method for pushing audio content according to claim 5, characterized in that: Also includes: Dynamically generate custom audio based on the health status and / or behavior data, wherein the custom audio includes generating symbolic music through text description, generating different emotional music, generating lyrics and melodies of different musical forms; and push the custom audio to the user.

12. An audio content push system, characterized in that: include: An EEG device module, which is used to collect EEG signals of the user; A data analysis module, which is used to identify the health status of the user based on the EEG signal; According to the health status, matching is performed with audio content in the audio content set, and calculating a first matching degree between each of the audio content and the health status; A treatment plan module, which is used to push the audio content to the user according to the first matching degree; According to the health status, the pushed audio content is changed until the health status reaches a preset value.

13. The audio content push system according to claim 12, characterized in that: Also includes: A music generation module is used to dynamically generate custom audio based on the health status and / or behavior data, wherein the custom audio includes generating symbolic music through text description, generating different emotional music, generating lyrics and melodies of different musical forms; and pushing the custom audio to users.

14. The audio content push system according to claim 12, characterized in that: The EEG device module is further configured as follows: Collecting at least one of the EEG signals of alpha wave, beta wave, gamma wave, theta wave and delta wave from the user's brain; At least one EEG parameter related to concentration, stress, emotion, fatigue, relaxation and mindfulness is calculated based on at least one of the EEG signals.

15. The audio content push system according to claim 12, characterized in that: The treatment plan module is further configured to: Acquire user behavior data on the audio content set; Based on the behavior data, calculating a second matching degree between each audio content in the audio content set and the user; Based on the first matching degree and / or the second matching degree, the audio content is pushed to the user.

Images