Music screening method and system for music therapy

Abstract

The present invention belongs to the technical field of brain wave data processing, and provides a music screening method and system for music therapy, which comprises the following steps: constructing and training an LSTM network as a stimulus recall interference elimination model; constructing and training an LSTM network as a music screening model by using the output result of the stimulus recall interference elimination model as a label; screening music through the music screening model to obtain music for music therapy; accurately identifying music imagination and recall segments in brain wave data, and eliminating signal interference caused by activation of brain regions during music imagination and recall, and not only eliminating strong interference generated by EEG brain wave signals generated by respective unique activation regions in the brain regions under audio stimulation, but also calculating the intensity of superposition-enhanced interference generated by overlapping, thereby intelligently eliminating interference of some potential value signals in the EEG brain wave signals of the subject under superposition-enhanced interference.

Description

Technical Field

[0001] The present invention belongs to the technical field of brain wave data processing, and in particular relates to a music screening method and system for music therapy. Background Art

[0002] At present, the therapeutic mechanism of music therapy is generally to make the inherent vibration frequency of the human body (such as heart rate, breathing, pulse, etc.) physically resonate with the music through the audio bpm in some specific music, thereby changing the response of the autonomic nervous system in the thalamus to stimulation so that music stimulation induces emotions, and can also regulate and improve emotions. The situation faced by each patient is different. Some patients need stimulating music to stimulate the cerebral cortex, while some patients need soothing music to soothe their emotions. Different patients need adapted music to resonate in order to adapt to music therapy. Therefore, screening the appropriate music is very critical for music therapy.

[0003] The current existing music screening method for music therapy is to collect the EEG signal of the user under the audio stimulation of music through EEG, and then screen the audio suitable for the user's emotions based on the EEG signal. For example, China Invention Publication No. CN111466908A discloses a method of using EEG to screen audio that can affect appetite. By analyzing the EEG by the first LSTM neural network, when training the second LSTM neural network, the trouble of collecting the appetite data of the test subject is eliminated. The brain waves of the test subject can be directly collected and put into the first LSTM neural network for analysis to obtain the appetite data of the test subject. This greatly speeds up the speed of generating training data for the second LSTM neural network and improves the efficiency of screening audio that affects appetite. However, when the EEG acquisition device collects the EEG signals of the user under the audio stimulation of music, it will simultaneously trigger the music listening process and the music imagination and recall process. The music imagination and recall process will activate multiple brain areas including the auditory cortex, frontal cortex, sensorimotor area, etc. The specific locations overlap with the brain areas activated by the music listening process, and there are also their own unique activation areas. See reference: Ding Y, Zhang Y, Zhou W, et al. Neural Correlates of Music Listening and Recall in the Human Brain[J]. The Journal of Neuroscience, 2019, 39(41): 1468-1418. DOI: 10.1523 / JNEUROSCI.1468-18.2019; therefore, the overlapping activated brain areas and their own unique activation areas generated by these music listening processes and music imagination and recall processes will have a strong interference with the accuracy of the music screened based on the user's emotions or cognitive judgments under the audio stimulation state. Summary of the invention

[0004] The purpose of the present invention is to propose a music screening method and system for music therapy to solve one or more technical problems existing in the prior art and at least provide a beneficial choice or create conditions.

[0005] In order to achieve the above object, according to one aspect of the present invention, a music screening method for music therapy is provided, and the music screening method for music therapy comprises the following steps:

[0006] S100, construct and train an LSTM network as a stimulus-recall interference elimination model;

[0007] S200, using the output of the stimulus recall interference elimination model as a label to build and train an LSTM network as a music screening model;

[0008] S300, filtering music through a music filtering model to obtain music for music therapy.

[0009] Further, in S100, the method for constructing and training an LSTM network as a stimulus recall interference elimination model is:

[0010] Build an LSTM network as the first network to collect the EEG data of the subjects in real time, switch the time slice every preset time, and randomly play different music to the subjects in different time slices. The subjects will score according to their feelings about the music.

[0011] Identify the music imagination and recall segments in the EEG data of each time slice, and eliminate the recall interference of the potential value of each music imagination and recall segment;

[0012] The EEG data after the recall interference is eliminated corresponding to each time slice is used as the input of the first network, and the scores corresponding to each time slice are used as corresponding labels to train the first network. The trained first network is recorded as the stimulus recall interference elimination model.

[0013] Further, in S200, the method of obtaining the music screening model by using the output result of the stimulus recall interference elimination model as a label to construct and train an LSTM network is as follows:

[0014] Build an LSTM network as the second network, collect the EEG data of the subjects in real time, switch a time slice at a preset time, play different music audios in different time slices as training audios, make a one-to-one correspondence between the audios in the same time slice and the collected EEG data, and input each EEG signal into the stimulus recall interference elimination model;

[0015] The audio of different music is input into the second network, and the output results of each EEG signal input into the stimulus-recall interference elimination model are used as labels of the second network. The trained second network is recorded as a music screening model.

[0016] The training audio is audio of music with a score ranging from eight to ten, which is scored by the subjects based on their own feelings about the music.

[0017] Preferably, the scoring value ranges from one to ten, with one representing the worst comfort and ten representing the best comfort.

[0018] After playing music, if the subject encounters music that is familiar to him or has been heard before, it will trigger the music listening process and the music imagination and recall process at the same time. The overlapping activated brain areas and the unique activated areas of the subjects will cause strong interference to the EEG brain wave signals generated under the audio stimulation state. Although the subjects are familiar with these music, some of them are not suitable for music therapy or will cause discomfort to the subjects. However, the common dual neural network recognition method in the prior art cannot identify this kind of music. For this reason, the present application provides the following scheme to identify the music imagination and recall segment in the EEG data and eliminate the signal interference caused by the activated brain areas during the music imagination and recall process:

[0019] Furthermore, the method for identifying the music imagination and recall segments in the EEG data of each time slice is as follows:

[0020] Perform artifact removal on EEG data;

[0021] Mark multiple sampling points on the EEG data after artifact removal according to the step length;

[0022] The serial number of the sampling point is recorded as i, and Epo(i) is the potential value of the i-th sampling point on the EEG signal of the EEG data; within the range of i, each Epo(i) is judged in turn, and the specific judgment steps are as follows:

[0023] The power spectral density of the data segment between the 1st sampling point to the i-th sampling point on the EEG signal is recorded as the reference spectral density of the reference Epo(i);

[0024] If the reference spectral density of Epo(i) is less than the reference spectral density of Epo(i-1) and the value of Epo(i) is less than the value of Epo(i-1), or, if the reference spectral density of Epo(i) is greater than the reference spectral density of Epo(i+1) and the value of Epo(i) is greater than the value of Epo(i+1), and at the same time, the value of Epo(i) is greater than EpoMid, then the sampling point corresponding to the EEG signal of the EEG data marked with Epo(i) is the imaginary recall starting punctuation ST;

[0025] If the reference spectral density of Epo(i) is greater than the reference spectral density of Epo(i-1) and the value of Epo(i) is greater than the value of Epo(i-1), or, if the reference spectral density of Epo(i) is less than the reference spectral density of Epo(i+1), the value of Epo(i) is less than the value of Epo(i+1), and at the same time, the value of Epo(i) is less than EpoMid, then the sampling point corresponding to the EEG signal of the EEG data marked with Epo(i) is the end point of imagination recall ET; wherein; EpoMid is the average value of the potential values ​​of all sampling points;

[0026] Each ST point is matched with the ET point closest to the ST point on the EEG signal in turn, and the data segment between each matched ST point and ET point on the EEG signal is recorded as a music imagination recall segment.

[0027] Preferably, the method for removing artifacts from EEG data is to remove artifacts from EEG data through time domain signal processing, and the time domain signal processing includes principal component analysis (PCA) or independent component analysis (ICA), wherein the independent component analysis (ICA) algorithm is: any one of the FastICA algorithm, JADE algorithm, extended maximum entropy algorithm and informax algorithm.

[0028] Preferably, the step length is 128 to 512 sampling points.

[0029] Furthermore, the method for eliminating the recall interference of the potential value of each music imagination recall segment is as follows:

[0030] The maximum potential value of the EEG signal in the music imagination and recall segment is IoBig, and the minimum potential value is IoSma;

[0031] The serial number of the potential value in the music imagination and recall segment is recorded as j, and Io(j) is the jth potential value in the music imagination and recall segment;

[0032] Eliminate the recall interference of each Io(j) in the music imagination recall segment, specifically:

[0033] The average of the potential value of the ST point corresponding to the music imagination and recall segment and IoBig is SMid;

[0034] The average value of the potential value of the ET point corresponding to the music imagination and recall segment and IoSma is EMid;

[0035] If Io(j)≤Smid,

[0036] Then Io(j)'=Sma(Io(j-1),Io(j),Io(j+1))+|SMid-Big(Io(j-1),Io(j),Io(j+1))|;

[0037] If Io(j)≥EMid,

[0038] Then Io(j)'=Big(Io(j-1),Io(j),Io(j+1))-|EMid-Sma(Io(j-1),Io(j),Io(j+1))|;

[0039] Io(j)' is the potential value after eliminating the recall interference of Io(j) in the music imagination recall segment;

[0040] Assign the potential value corresponding to Io(j) in the music imagination and recall segment to Io(j)';

[0041] Sma(Io(j-1), Io(j), Io(j+1)) is the smallest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1);

[0042] Big(Io(j-1),Io(j),Io(j+1)) is the largest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1).

[0043] Although the above method can eliminate partial interference of some potential value signals of the subjects in the process of music imagination and recall, it can only eliminate the strong interference of EEG brain wave signals generated by the unique activation areas in the brain under the audio stimulation state by directly comparing the adjacent values. However, if multiple brain areas including the auditory cortex, frontal cortex, sensorimotor area, etc. are activated in the process of music imagination and recall, and there is overlap, the elimination strength is not enough to completely remove the superposition-enhanced interference generated by the overlap. For this reason, the present application proposes the following method to calculate the intensity of the superposition-enhanced interference generated by the overlap, and then intelligently eliminate the interference of some potential value signals of the subjects under the superposition-enhanced interference:

[0044] Preferably, the method of eliminating the recall interference of the potential value of each music imagination recall segment is replaced by:

[0045] The maximum potential value of the EEG signal in the music imagination and recall segment is IoBig, and the minimum potential value is IoSma;

[0046] The serial number of the potential value in the music imagination and recall segment is recorded as j, and Io(j) is the jth potential value in the music imagination and recall segment;

[0047] Eliminate the recall interference of each Io(j) in the music imagination recall segment, specifically:

[0048] The average of the potential value of the ST point corresponding to the music imagination and recall segment and IoBig is SMid;

[0049] The average value of the potential value of the ET point corresponding to the music imagination and recall segment and IoSma is EMid;

[0050] If Io(j)≤Smid, then Io(j)'=Sma(Io(j-1),Io(j),Io(j+1))+Recall(j);

[0051] If Io(j)≥EMid, then Io(j)'=Big(Io(j-1),Io(j),Io(j+1))-Recall(j);

[0052] Io(j)' is the potential value after eliminating the recall interference of Io(j) in the music imagination recall segment;

[0053] Assign the potential value corresponding to Io(j) in the music imagination and recall segment to Io(j)';

[0054] Sma(Io(j-1), Io(j), Io(j+1)) is the smallest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1);

[0055] Big(Io(j-1),Io(j),Io(j+1)) is the largest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1).

[0056] Among them, Recall(j) is the superposition interference intensity of Io(j) in the music imagination recall segment, and its calculation method is:

[0057] ;

[0058] Where IM is the total number of potential values ​​in the music imagination recall segment, k is a variable, Big(Io(k), Io(j)) is the larger value between Io(k) and Io(j), and Sma(Io(j), Io(k)) is the smaller value between Io(j) and Io(k).

[0059] Furthermore, the time slice is a time period of 5 seconds to 60 seconds.

[0060] Furthermore, the preset time is from 5 seconds to 60 seconds.

[0061] Furthermore, the EEG signal is input at a sampling rate of 200 Hz or above.

[0062] Furthermore, the method for collecting the EEG data of the subject is to collect the EEG data of the subject through a BrainAmp EEG signal collector or a SynAmps Model 8050 EEG signal collector.

[0063] Furthermore, in S300, the method for obtaining music for music therapy by screening music through a music screening model is as follows: different music is used as music to be screened, the music to be screened is input into the music screening model, and whether it is music for music therapy of the subject is determined according to the output result of the music screening model.

[0064] The present invention also provides a music screening system for music therapy, the system comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to run in the following system units:

[0065] Recall elimination modeling unit, used to build and train LSTM network as stimulus recall interference elimination model;

[0066] A music screening modeling unit is used to construct and train an LSTM network as a music screening model using the output of the stimulus recall interference elimination model as a label;

[0067] The therapeutic music screening unit is used to screen the music through a music screening model to obtain music for music therapy.

[0068] The beneficial effects of the present invention are as follows: the present invention provides a music screening method and system for music therapy, which can accurately identify the music imagination and recall segments in the EEG data, and eliminate the signal interference caused by the activation of the brain area during the music imagination and recall process, and can not only eliminate the strong interference caused by the EEG electroencephalogram signals generated by the unique activation areas in the brain areas under the audio stimulation state, but also calculate the intensity of the superimposed interference generated by the overlapping type, thereby intelligently eliminating the interference of some potential value signals in the EEG electroencephalogram signals of the subjects under the superimposed interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The above and other features of the present invention will become more obvious by describing in detail the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar elements. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings:

[0070] Figure 1 Shown is a flow chart of a music screening method for music therapy;

[0071] Figure 2 Shown is a structural diagram of a music screening system for music therapy. DETAILED DESCRIPTION

[0072] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0073] Example 1

[0074] like Figure 1 The flowchart of a music screening method for music therapy in Example 1 is shown below. Figure 1A music screening method for music therapy according to embodiment 1 of the present invention is described, and the method comprises the following steps:

[0075] S100, construct and train an LSTM network as a stimulus recall interference elimination model;

[0076] S200, using the output of the stimulus recall interference elimination model as a label to build and train an LSTM network as a music screening model;

[0077] S300, filtering music through a music filtering model to obtain music for music therapy.

[0078] Further, in S100, the method for constructing and training an LSTM network as a stimulus recall interference elimination model is:

[0079] Build an LSTM network as the first network to collect the EEG data of the subjects in real time, switch the time slice every preset time, and randomly play different music to the subjects in different time slices. The subjects will score according to their feelings about the music.

[0080] Identify the music imagination and recall segments in the EEG data of each time slice, and eliminate the recall interference of the potential value of each music imagination and recall segment;

[0081] The EEG data after the recall interference is eliminated corresponding to each time slice is used as the input of the first network, and the scores corresponding to each time slice are used as corresponding labels to train the first network. The trained first network is recorded as the stimulus recall interference elimination model.

[0082] Further, in S200, the method of obtaining the music screening model by using the output result of the stimulus recall interference elimination model as a label to construct and train an LSTM network is as follows:

[0083] Build an LSTM network as the second network, collect the EEG data of the subjects in real time, switch a time slice at a preset time, play different music audios in different time slices as training audios, make a one-to-one correspondence between the audios in the same time slice and the collected EEG data, and input each EEG signal into the stimulus recall interference elimination model;

[0084] The audio of different music is input into the second network, and the output results of each EEG signal input into the stimulus-recall interference elimination model are used as labels of the second network. The trained second network is recorded as a music screening model.

[0085] The training audio is audio of music with a score ranging from eight to ten, which is scored by the subjects based on their own feelings about the music.

[0086] Preferably, the scoring value ranges from one to ten, with one representing the worst comfort and ten representing the best comfort.

[0087] Furthermore, the method for identifying the music imagination and recall segments in the EEG data of each time slice is as follows:

[0088] Perform artifact removal on EEG data;

[0089] Mark multiple sampling points on the EEG data after artifact removal according to the step length;

[0090] The serial number of the sampling point is recorded as i, and Epo(i) is the potential value of the i-th sampling point on the EEG signal of the EEG data; within the range of i, each Epo(i) is judged in turn, and the specific judgment steps are as follows:

[0091] The power spectral density of the data segment between the 1st sampling point to the i-th sampling point on the EEG signal is recorded as the reference spectral density of the reference Epo(i);

[0092] If the reference spectral density of Epo(i) is less than the reference spectral density of Epo(i-1) and the value of Epo(i) is less than the value of Epo(i-1), or, if the reference spectral density of Epo(i) is greater than the reference spectral density of Epo(i+1) and the value of Epo(i) is greater than the value of Epo(i+1), and at the same time, the value of Epo(i) is greater than EpoMid, then the sampling point corresponding to the EEG signal of the EEG data marked with Epo(i) is the imaginary recall starting punctuation ST;

[0093] If the reference spectral density of Epo(i) is greater than the reference spectral density of Epo(i-1) and the value of Epo(i) is greater than the value of Epo(i-1), or, if the reference spectral density of Epo(i) is less than the reference spectral density of Epo(i+1), the value of Epo(i) is less than the value of Epo(i+1), and at the same time, the value of Epo(i) is less than EpoMid, then the sampling point corresponding to the EEG signal of the EEG data marked with Epo(i) is the end point of imagination recall ET; wherein; EpoMid is the average value of the potential values ​​of all sampling points;

[0094] Each ST point is matched with the ET point closest to the ST point on the EEG signal in turn, and the data segment between each matched ST point and ET point on the EEG signal is recorded as a music imagination recall segment.

[0095] Among them, the method for removing artifacts from EEG data is to remove artifacts from EEG data through time domain signal processing, and the time domain signal processing is the FastICA algorithm.

[0096] Preferably, the step length is 512 sampling points.

[0097] Furthermore, the method for eliminating the recall interference of the potential value of each music imagination recall segment is as follows:

[0098] The maximum potential value of the EEG signal in the music imagination and recall segment is IoBig, and the minimum potential value is IoSma;

[0099] The serial number of the potential value in the music imagination and recall segment is recorded as j, and Io(j) is the jth potential value in the music imagination and recall segment;

[0100] Eliminate the recall interference of each Io(j) in the music imagination recall segment, specifically:

[0101] The average of the potential value of the ST point corresponding to the music imagination and recall segment and IoBig is SMid;

[0102] The average value of the potential value of the ET point corresponding to the music imagination and recall segment and IoSma is EMid;

[0103] If Io(j)≤Smid,

[0104] Then Io(j)'=Sma(Io(j-1),Io(j),Io(j+1))+|SMid-Big(Io(j-1),Io(j),Io(j+1))|;

[0105] If Io(j)≥EMid,

[0106] Then Io(j)'=Big(Io(j-1),Io(j),Io(j+1))-|EMid-Sma(Io(j-1),Io(j),Io(j+1))|;

[0107] Io(j)' is the potential value after eliminating the recall interference of Io(j) in the music imagination recall segment;

[0108] Assign the potential value corresponding to Io(j) in the music imagination and recall segment to Io(j)';

[0109] Sma(Io(j-1), Io(j), Io(j+1)) is the smallest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1);

[0110] Big(Io(j-1),Io(j),Io(j+1)) is the largest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1).

[0111] Furthermore, the time slice is a 10-second time period.

[0112] Furthermore, the preset time is 10 seconds.

[0113] Furthermore, the EEG signal is input at a sampling rate of 200 Hz.

[0114] Furthermore, the method for collecting the EEG data of the subject is to collect the EEG data of the subject through a BrainAmp EEG signal collector.

[0115] Further, in S300, the method for obtaining music for music therapy by screening music through a music screening model is as follows: different music is used as music to be screened, the music to be screened is input into the music screening model, and whether it is music for music therapy of the subject is determined according to the output result of the music screening model, that is, if the music to be screened passes the screening result of the music screening model as a positive sample, the music to be screened is marked as music for music therapy; if the music to be screened passes the screening result of the music screening model as a negative sample, the music to be screened is marked as not music for music therapy.

[0116] Example 2

[0117] This embodiment is to replace the method of eliminating the recall interference of the potential value of each music imagination recall segment in the embodiment 1 with the following, specifically:

[0118] Preferably, the method of eliminating the recall interference of the potential value of each music imagination recall segment is replaced by:

[0119] The maximum potential value of the EEG signal in the music imagination and recall segment is IoBig, and the minimum potential value is IoSma;

[0120] The serial number of the potential value in the music imagination and recall segment is recorded as j, and Io(j) is the jth potential value in the music imagination and recall segment;

[0121] Eliminate the recall interference of each Io(j) in the music imagination recall segment, specifically:

[0122] The average of the potential value of the ST point corresponding to the music imagination and recall segment and IoBig is SMid;

[0123] The average value of the potential value of the ET point corresponding to the music imagination and recall segment and IoSma is EMid;

[0124] If Io(j)≤Smid, then Io(j)'=Sma(Io(j-1),Io(j),Io(j+1))+Recall(j);

[0125] If Io(j)≥EMid, then Io(j)'=Big(Io(j-1),Io(j),Io(j+1))-Recall(j);

[0126] Io(j)' is the potential value after eliminating the recall interference of Io(j) in the music imagination recall segment;

[0127] Assign the potential value corresponding to Io(j) in the music imagination and recall segment to Io(j)';

[0128] Sma(Io(j-1), Io(j), Io(j+1)) is the smallest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1);

[0129] Big(Io(j-1),Io(j),Io(j+1)) is the largest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1).

[0130] Among them, Recall(j) is the superposition interference intensity of Io(j) in the music imagination recall segment, and its calculation method is:

[0131] ;

[0132] Where IM is the total number of potential values ​​in the music imagination recall segment, k is a variable, Big(Io(k), Io(j)) is the larger value of Io(k) and Io(j), and Sma(Io(j), Io(k)) is the smaller value of Io(j) and Io(k).

[0133] In addition, the present invention also provides an embodiment of a music screening system for music therapy, such as Figure 2 Shown is a structural diagram of a music screening system for music therapy of the present invention. The music screening system for music therapy in this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned embodiment of the music screening system for music therapy are implemented.

[0134] The system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to run in the following units of the system:

[0135] Recall elimination modeling unit, used to build and train LSTM network as stimulus recall interference elimination model;

[0136] A music screening modeling unit is used to construct and train an LSTM network as a music screening model using the output of the stimulus recall interference elimination model as a label;

[0137] The therapeutic music screening unit is used to screen the music through a music screening model to obtain music for music therapy.

[0138] The music screening system for music therapy can be run on computing devices such as desktop computers, notebooks, PDAs, and cloud servers. The music screening system for music therapy can include, but is not limited to, processors and memories. Those skilled in the art will appreciate that the example is merely an example of a music screening system for music therapy and does not constitute a limitation on a music screening system for music therapy. It can include more or fewer components than the example, or a combination of certain components, or different components. For example, the music screening system for music therapy can also include input and output devices, network access devices, buses, etc.

[0139] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DEPO), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the music screening system for music therapy, and uses various interfaces and lines to connect the various parts of the entire music screening system for music therapy.

[0140] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the music screening system for music therapy by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0141] Although the description of the present invention has been quite detailed and has been described in particular with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present invention. In addition, the present invention is described above with the embodiments foreseeable by the inventors, and its purpose is to provide a useful description, and those non-substantial changes to the present invention that are not currently foreseen may still represent equivalent changes of the present invention.

Claims

1. A music screening method for music therapy, characterized in that: The method comprises the following steps: S100, construct and train an LSTM network as a stimulus-recall interference elimination model; S200, using the output of the stimulus recall interference elimination model as a label to build and train an LSTM network as a music screening model; S300, filtering music through a music screening model to obtain music for music therapy; Among them, the method for constructing and training the LSTM network as a stimulus recall interference elimination model is: Build an LSTM network as the first network to collect the EEG data of the subjects in real time, switch the time slice every preset time, and randomly play different music to the subjects in different time slices. The subjects will score according to their feelings about the music. Identify the music imagination and recall segments in the EEG data of each time slice, and eliminate the recall interference of the potential value of each music imagination and recall segment; The EEG data after the recall interference elimination corresponding to each time slice is used as the input of the first network, the scores corresponding to each time slice are used as the corresponding labels to train the first network, and the trained first network is recorded as the stimulus recall interference elimination model; In S200, the output result of the stimulus recall interference elimination model is used as a label to construct and train an LSTM network as a music screening model. The method is as follows: Build an LSTM network as the second network, collect the EEG data of the subjects in real time, switch a time slice at a preset time, play different music audios in different time slices as training audios, make a one-to-one correspondence between the audios in the same time slice and the collected EEG data, and input each EEG signal into the stimulus recall interference elimination model; The audio of different music is input into the second network, and the output results of each EEG signal input into the stimulus recall interference elimination model are used as the label of the second network, and the trained second network is recorded as the music screening model; The method for identifying the music imagination and recall segments in the EEG data of each time slice is as follows: Perform artifact removal on EEG data; Mark multiple sampling points on the EEG data after artifact removal according to the step length; The serial number of the sampling point is recorded as i, and Epo(i) is the potential value of the i-th sampling point on the EEG signal of the EEG data; within the range of i, each Epo(i) is judged in turn, and the specific judgment steps are as follows: The power spectral density of the data segment between the 1st sampling point to the i-th sampling point on the EEG signal is recorded as the reference spectral density of the reference Epo(i); If the reference spectral density of Epo(i) is less than the reference spectral density of Epo(i-1) and the value of Epo(i) is less than the value of Epo(i-1), or, if the reference spectral density of Epo(i) is greater than the reference spectral density of Epo(i+1) and the value of Epo(i) is greater than the value of Epo(i+1), and at the same time, the value of Epo(i) is greater than EpoMid, then the sampling point corresponding to the EEG signal of the EEG data marked with Epo(i) is the imaginary recall starting punctuation ST; If the reference spectral density of Epo(i) is greater than the reference spectral density of Epo(i-1) and the value of Epo(i) is greater than the value of Epo(i-1), or, if the reference spectral density of Epo(i) is less than the reference spectral density of Epo(i+1), the value of Epo(i) is less than the value of Epo(i+1), and at the same time, the value of Epo(i) is less than EpoMid, then the sampling point corresponding to the EEG signal of the EEG data marked with Epo(i) is the end point of imagination recall ET; wherein; EpoMid is the average value of the potential values ​​of all sampling points; Sequentially match each ST point with the ET point closest to the ST point on the EEG signal, and record the data segment between each matched ST point and ET point on the EEG signal as a music imagination recall segment; Among them, the method for eliminating the recall interference of the potential value of each music imagination recall segment is: The maximum potential value of the EEG signal in the music imagination and recall segment is IoBig, and the minimum potential value is IoSma; The serial number of the potential value in the music imagination and recall segment is recorded as j, and Io(j) is the jth potential value in the music imagination and recall segment; Eliminate the recall interference of each Io(j) in the music imagination recall segment, specifically: The average of the potential value of the ST point corresponding to the music imagination and recall segment and IoBig is SMid; The average value of the potential value of the ET point corresponding to the music imagination and recall segment and IoSma is EMid; If Io(j)≤Smid, Then Io(j)'=Sma(Io(j-1),Io(j),Io(j+1))+|SMid-Big(Io(j-1),Io(j),Io(j+1))|; If Io(j)≥EMid, Then Io(j)'=Big(Io(j-1),Io(j),Io(j+1))-|EMid-Sma(Io(j-1),Io(j),Io(j+1))|; Io(j)' is the potential value after eliminating the recall interference of Io(j) in the music imagination recall segment; Assign the potential value corresponding to Io(j) in the music imagination and recall segment to Io(j)'; Sma(Io(j-1), Io(j), Io(j+1)) is the smallest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1); Big(Io(j-1),Io(j),Io(j+1)) is the largest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1).

2. A music screening method for music therapy according to claim 1, characterized in that: The method of eliminating the recall interference of the potential value of each music imagination recall segment is replaced by: The maximum potential value of the EEG signal in the music imagination and recall segment is IoBig, and the minimum potential value is IoSma; The serial number of the potential value in the music imagination and recall segment is recorded as j, and Io(j) is the jth potential value in the music imagination and recall segment; Eliminate the recall interference of each Io(j) in the music imagination recall segment, specifically: The average of the potential value of the ST point corresponding to the music imagination and recall segment and IoBig is SMid; The average value of the potential value of the ET point corresponding to the music imagination and recall segment and IoSma is EMid; If Io(j)≤Smid, then Io(j)'=Sma(Io(j-1),Io(j),Io(j+1))+Recall(j); If Io(j)≥EMid, then Io(j)'=Big(Io(j-1),Io(j),Io(j+1))-Recall(j); Io(j)' is the potential value after eliminating the recall interference of Io(j) in the music imagination recall segment; Assign the potential value corresponding to Io(j) in the music imagination and recall segment to Io(j)'; Sma(Io(j-1), Io(j), Io(j+1)) is the smallest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1); Big(Io(j-1),Io(j),Io(j+1)) is the largest potential value among the three values ​​of Io(j), Io(j-1) and Io(j+1); Recall(j) is the superimposed interference intensity of Io(j) in the music imagination recall segment.

3. A music screening method for music therapy according to claim 2, characterized in that: Recall(j) is calculated as: ； Where IM is the total number of potential values ​​in the music imagination recall segment, k is a variable, Big(Io(k), Io(j)) is the larger value between Io(k) and Io(j), and Sma(Io(j), Io(k)) is the smaller value between Io(j) and Io(k).

4. The music screening method for music therapy according to claim 1, characterized in that: The method for collecting the EEG data of the subject is to collect the EEG data of the subject through a BrainAmp EEG signal collector or a SynAmps Model 8050 EEG signal collector.

5. The music screening method for music therapy according to claim 1, characterized in that: In S300, the method for obtaining music for music therapy by screening music through a music screening model is as follows: different music is used as music to be screened, the music to be screened is input into the music screening model, and whether it is music for music therapy of the subject is determined according to the output result of the music screening model.

6. A music screening system for music therapy, characterized in that: The music screening system for music therapy comprises: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the music screening method for music therapy described in any one of claims 1 to 5 are implemented.

Images