Music-induced emotion recognition method and system based on EEG

Abstract

The present invention relates to the field of emotion recognition technology, and discloses a music-induced emotion recognition method and system based on EEG, including the following steps: S11, designing a BCI experimental paradigm based on music-stimulated emotions, wherein the music stimulation includes two types, the joyful and happy type and the sad and painful type; S12, using E‑prime to design the experimental task interface, including the gaze interface, the music stimulation interface, the feedback interface, and the rest interface; before the experiment begins, the center of the screen will display instructions to let the subjects fully understand the experimental rules, and before the formal data collection, the subjects adjust their emotions and are in a state of physical and mental relaxation, and wear headphones to prepare for the experiment. The music-induced emotion recognition method and system based on EEG designs a brain-computer interface experimental paradigm for music-stimulated emotions. Compared with other ways of stimulating emotions, music stimulation can make the subjects more relaxed and stable, which is more conducive to emotional treatment and has a good prospect for the treatment of depression.

Description

Technical Field

[0001] The present invention relates to the technical field of emotion recognition, and in particular to an EEG-based music-induced emotion recognition method and system. Background Art

[0002] Emotional brain-computer interfaces reflect human cognition and attitudes to a large extent. With the development of brain science and psychology, emotion recognition has gradually been valued by people and has become one of the research focuses in this field. It is particularly important to know the emotions of certain patients in clinical practice, such as patients with depression and emotional disorders. At the same time, for patients who have lost their language ability, detecting their emotional state is also an important channel. In fact, the application scope of emotion research is far more than that. Emotional task research can be used for the education of special children and understanding the psychology of children with autism. In business life, such as the development of software products, user satisfaction with software products is crucial. Traditional market research is to distribute questionnaires to a large number of users, but users often do not take it seriously, so most of the results obtained cannot truly reflect the quality of the product. Using an electroencephalogram to collect EEG data when users use the product, and then analyzing the emotional state, is more objective and true. Similarly, emotional assessment of some practitioners is also an important application in the field of emotion research, such as soldiers, pilots, long-distance drivers, etc.

[0003] Existing emotion induction methods include spontaneous recall, picture emotion induction, and video clip induction. In the spontaneous recall method, the subject closes his eyes and will generate emotions autonomously when recalling the present moment because he is immersed in the event, which is difficult to achieve by relying solely on the subject's own memories. The picture emotion induction method mainly allows the subject to watch pictures with strong emotional colors to induce the required target emotional state. Compared with the picture-induced emotion experiment, each picture is displayed for only a few seconds, and the emotion induction time is very short. Therefore, the induced emotions are short-lived, and it is difficult to obtain a certain sustained and stable emotion from the subject, which has certain limitations. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides an EEG-based music-induced emotion recognition method and system, and designs a brain-computer interface experimental paradigm for music-stimulated emotions. Compared with other ways of stimulating emotions, music stimulation can put the subjects in a relaxed and stable state, which is more conducive to emotional treatment and has good prospects for the treatment of depression. The subjects can enhance their emotions by associating different types of music with their own experiences of pleasure or sadness. When listening to music, the subjects are in a state of complete relaxation, which reduces the interference caused by eye movements.

[0006] (II) Technical solution

[0007] The present invention provides the following technical solution: a method and system for identifying music-induced emotions based on EEG, comprising the following steps:

[0008] S1, designed the experimental paradigm and collected EEG data.

[0009] S11. Design a BCI experimental paradigm based on music stimulation emotions. Music stimulation includes two types, joyful and happy type and sad and heartbreaking type.

[0010] S12. Use E-prime to design the experimental task interface, including the gaze interface, music stimulation interface, feedback interface, and rest interface.

[0011] S13. Before the experiment begins, instructions will be displayed in the center of the screen to allow the subjects to fully understand the experimental rules. Before the formal data collection, the subjects should adjust their emotions and be in a state of physical and mental relaxation, wear headphones and be ready for the experiment.

[0012] S14. At the beginning of the experiment, a plus sign "+" will appear in the center of the screen for 3 seconds, prompting the subjects to prepare themselves for the mood. Then, when the music symbol appears, the music will start playing. The music will play for 60 seconds, and then the screen will prompt "Please rate the type of music" for 5 seconds. If the subject thinks the music is positive, press the "9" key; if he thinks the music is negative, press the "1" key; if he cannot judge, press the "5" key. Finally, an asterisk "*" will be displayed in the center of the screen for 2 seconds. During this period, the subjects will rest and adjust their emotions.

[0013] S2. Preprocess the collected EEG signals.

[0014] S21. Three sets of EEG data of the subjects are obtained by measuring electrodes. Joyful and cheerful music and sad and painful music are sorted into two data sets based on the moment label data. First, the original EEG signal is processed using a 50 Hz notch filter.

[0015] S22. Use a 6th-order Butterworth filter to filter the EEG signal in the range of 1-50 Hz. According to the time labels of the measuring electrodes, the EEG data is segmented and processed. The 60-second time period of music playback is taken for processing. Then, time-frequency analysis is performed on all electrodes. The power spectral density diagrams of the 16 electrodes of the two types of data and the power spectral density difference diagrams of the two types of music data are drawn.

[0016] S23. Based on the power spectrum density diagram and the difference diagram of the power spectra of the two types of emotions, the difference changes of the two types of emotions at electrode P4 in the frequency band of 15-25Hz were obtained. The 9 positive emotion experimental data and 9 negative emotion experimental data of electrode P4 at the superimposed frequency of 15-25Hz were taken as two samples, and a paired t-test was performed on the two samples, and P<0.05 was obtained, proving the difference between the two types of emotions.

[0017] S24, consider adjusting the bandpass filter range to 15-25Hz, after determining the frequency range of energy change, set the data energy in the 60-second music playback after filtering to A, set the energy of the reference time period before the prompt to R, and the energy fluctuation percentage D of the 60-second data relative to the energy before the prompt occurs to

[0018]

[0019] S3. Take 60 seconds of music playing time for feature extraction.

[0020] S31. Convert the time domain information into time-frequency information, perform short-time Fourier transform on each group of each guide signal, use Hanning window as the window function, and set the window size to 1 second, the time domain sampling frequency to 250Hz, the frequency domain sampling frequency to 512Hz, and the emotional information contained in the characteristics of each frequency band is different. In this way, the time domain information within 1 second will be converted to the frequency domain through a Fourier transform. In this way, for each type of emotional data, 9*60*250=135,000 sampling points are obtained.

[0021] S32. Take out the five frequency bands closely related to physiological and psychological activities, namely Delta band 1-4Hz, Theta band 4-8Hz, Alpha band 8-13Hz, Beta band 13-30Hz and Gamma band 30-50Hz, extract differential entropy frequency domain features for a single frequency band to obtain 16 differential entropy (DE) features, extract differential entropy features for five frequency bands, and extract 16*5=80 differential entropy (DE) features for each group of data.

[0022] S4. Use PCA method to reduce the dimension of features.

[0023] S41. First, the data is decentralized and normalized by subtracting the mean value from each variable. This way, the center of the data can be returned to the zero point, that is, the mean is 0.

[0024] S42. Calculate the covariance matrix of the centered sample X.

[0025] S43. Calculate the eigenvalues ​​of the covariance matrix and arrange them in descending order.

[0026] S44. Calculate the eigenvector corresponding to the eigenvalue, and take the eigenvectors corresponding to the first p eigenvalues ​​to form a projection matrix.

[0027] S45. Convert the data into the new space constructed by the above projection matrix, so as to achieve the purpose of dimensionality reduction.

[0028] S5. The data after dimensionality reduction were divided into training set and test set according to ten-fold cross validation, and the data were classified by SVM to obtain the average classification accuracy of the six subjects in Delta, Theta, Alpha, Beta, Gamma and five frequency bands.

[0029] Preferably, in step S1, the EEG signal acquisition device comprises 16 electrodes: F3, FZ, F4, FC1, FC2, FC5, FC6, C3, CZ, C4, CP1, CP2, CP5, CP6, P3, PZ, P4, and the sampling rate is 250Hz. According to the international standard lead 10 / 20 system setting, the electrodes to be measured are placed on the scalp of the subject, the reference electrode and the ground electrode are located at the earlobe and forehead, and the electrode output end is connected to the wireless EEG acquisition system.

[0030] Preferably, in step S1, music of each type of emotion appears randomly three times in a group of experiments, and each subject completes three groups of experiments in total, with a twenty-minute rest between each group of experiments.

[0031] Preferably, in step S1, 6 1-minute music clips are used as stimulus materials, of which 3 clips reflect joyful and happy emotions, and 3 clips reflect sad and painful emotions. The music is played randomly, and there is a 5-second feedback time after each clip is played for 60 seconds. If the subject thinks that the music is positive and cheerful music, he presses the "9" key, and if he thinks that the music is negative music, he presses the "1" key. If he cannot judge the type of music, he presses the 5 key. Each subject is required to do 3 sets of experiments, with a 20-minute interval between each set of experiments.

[0032] The EEG-based music-induced emotion recognition system comprises a data acquisition unit 1 , wherein the output end signal of the data acquisition unit 1 is connected to a system control unit 2 , and the output end signal of the system control unit 2 is connected to a data processing unit 3 .

[0033] Preferably, the data acquisition unit 1 comprises an acquisition module 11 , the output end signal of the acquisition module 11 is connected to a receiving module 12 , the output end signal of the receiving module 12 is connected to an uploading module 13 , and the output end signal of the uploading module 13 is connected to a storage module 14 .

[0034] Preferably, the system control unit 2 includes a touch module 21 , the output end signal of the touch module 21 is connected to a sensor module 22 , the output end signal of the sensor module 22 is connected to a feedback module 23 , and the output end signal of the feedback module 23 is connected to an output module 24 .

[0035] Preferably, the data processing unit 3 comprises a calling module 31 , the output end signal of the calling module 31 is connected to an analysis module 32 , the output end signal of the analysis module 32 is connected to a processing module 33 , and the output end signal of the processing module 33 is connected to a display module 34 .

[0036] (III) Beneficial effects

[0037] Compared with the prior art, the present invention provides a music-induced emotion recognition method and system based on EEG, which has the following beneficial effects:

[0038] 1. This EEG-based music-induced emotion recognition method and system designs a brain-computer interface experimental paradigm for music-stimulated emotions. Compared with other ways of stimulating emotions, music stimulation can put the subjects in a more relaxed and stable state, which is more conducive to emotional treatment and has good prospects for the treatment of depression. The subjects can enhance their emotions by associating different types of music with their own experiences of pleasure or sadness.

[0039] 2. This EEG-based music-induced emotion recognition method and system allows the subject to be in a state of complete relaxation when listening to music, which reduces the interference caused by eye movements. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a diagram of the experimental paradigm of the EEG-based emotion recognition system of the present invention;

[0041] Figure 2 This is the distribution diagram of the positions of 16 electrodes of the present invention;

[0042] Figure 3 It is the power spectrum density diagram of positive music and negative music of the present invention;

[0043] Figure 4 It is a power spectrum density difference diagram of positive music and negative music of the present invention;

[0044] Figure 5 It is the energy fluctuation diagram of the positive music and negative music of the present invention;

[0045] Figure 6 This is a statistical chart of the average classification accuracy of the six subjects in different frequency bands of the present invention;

[0046] Figure 7 It is the principle diagram of the system of the present invention;

[0047] Figure 8 It is a schematic diagram of a data acquisition unit of the present invention;

[0048] Fig. 9 It is a schematic diagram of a control unit of the system of the present invention;

[0049] Fig.10 Schematic diagram of the data processing unit of the present invention.

[0050] In the accompanying drawings: 1. Data acquisition unit; 11. Acquisition module; 12. Receiving module; 13. Upload module; 14. Storage module; 2. System control unit; 21. Touch module; 22. Sensing module; 23. Feedback module; 24. Output module; 3. Data processing unit; 31. Calling module; 32. Analysis module; 33. Processing module; 34. Display module. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Embodiment 1:

[0053] See also Figure 1-6 The present invention provides a technical solution: a method and system for identifying music-induced emotions based on EEG, comprising the following steps:

[0054] S1, designed the experimental paradigm and collected EEG data.

[0055] S11. Design a BCI experimental paradigm based on music stimulation emotions. Music stimulation includes two types, joyful and happy type and sad and heartbreaking type.

[0056] S12. Use E-prime to design the experimental task interface, including the gaze interface, music stimulation interface, feedback interface, and rest interface.

[0057] S13. Before the experiment begins, instructions will be displayed in the center of the screen to allow the subjects to fully understand the experimental rules. Before the formal data collection, the subjects should adjust their emotions and be in a state of physical and mental relaxation, wear headphones and be ready for the experiment.

[0058] S14. At the beginning of the experiment, a plus sign "+" will appear in the center of the screen for 3 seconds, prompting the subjects to prepare themselves for the mood. Then, when the music symbol appears, the music will start playing. The music will play for 60 seconds, and then the screen will prompt "Please rate the type of music" for 5 seconds. If the subject thinks the music is positive, press the "9" key; if he thinks the music is negative, press the "1" key; if he cannot judge, press the "5" key. Finally, an asterisk "*" will be displayed in the center of the screen for 2 seconds. During this period, the subjects will rest and adjust their emotions.

[0059] The EEG signal acquisition equipment contains 16 electrodes: F3, FZ, F4, FC1, FC2, FC5, FC6, C3, CZ, C4, CP1, CP2, CP5, CP6, P3, PZ, P4, with a sampling rate of 250Hz. According to the international standard lead 10 / 20 system settings, the electrodes to be measured are placed on the scalp of the subjects, the reference electrode and the ground electrode are located at the earlobe and forehead, and the electrode output is connected to the wireless EEG acquisition system. Each type of emotional music appears randomly three times in a set of experiments. Each subject completes three sets of experiments in total and rests for 20 minutes between each set of experiments.

[0060] Six one-minute music clips were used as stimulus materials, of which three clips reflected joyful and happy emotions, and three clips reflected sad and painful emotions. The music was played randomly, and there was a 5-second feedback time after each clip was played for 60 seconds. If the subject thought the music was positive and cheerful, he would press the "9" key, if he thought the music was negative and sad, he would press the "1" key, and if he could not judge the type of music, he would press the 5 key. Each subject was asked to do three sets of experiments, with a twenty-minute interval between each set of experiments.

[0061] S2. Preprocess the collected EEG signals.

[0062] S21. Three sets of EEG data of the subjects are obtained by measuring electrodes. Joyful and cheerful music and sad and painful music are sorted into two data sets based on the moment label data. First, the original EEG signal is processed using a 50 Hz notch filter.

[0063] S22. Use a 6th-order Butterworth filter to filter the EEG signal in the range of 1-50 Hz. According to the time labels of the measuring electrodes, the EEG data is segmented and processed. The 60-second time period of music playback is taken for processing. Then, time-frequency analysis is performed on all electrodes. The power spectral density diagrams of the 16 electrodes of the two types of data and the power spectral density difference diagrams of the two types of music data are drawn.

[0064] S23. Based on the power spectrum density diagram and the difference diagram of the power spectra of the two types of emotions, the difference changes of the two types of emotions at electrode P4 in the frequency band of 15-25Hz were obtained. The 9 positive emotion experimental data and 9 negative emotion experimental data of electrode P4 at the superimposed frequency of 15-25Hz were taken as two samples, and a paired t-test was performed on the two samples, and P<0.05 was obtained, proving the difference between the two types of emotions.

[0065] S24, consider adjusting the bandpass filter range to 15-25Hz, after determining the frequency range of energy change, set the data energy in the 60-second music playback after filtering to A, set the energy of the reference time period before the prompt to R, and the energy fluctuation percentage D of the 60-second data relative to the energy before the prompt occurs to

[0066]

[0067] S3. Take 60 seconds of music playing time for feature extraction.

[0068] S31. Convert the time domain information into time-frequency information, perform short-time Fourier transform on each group of each guide signal, use Hanning window as the window function, and set the window size to 1 second, the time domain sampling frequency to 250Hz, the frequency domain sampling frequency to 512Hz, and the emotional information contained in the characteristics of each frequency band is different. In this way, the time domain information within 1 second will be converted to the frequency domain through a Fourier transform. In this way, for each type of emotional data, 9*60*250=135,000 sampling points are obtained.

[0069] S32. Take out the five frequency bands closely related to physiological and psychological activities, namely Delta band 1-4Hz, Theta band 4-8Hz, Alpha band 8-13Hz, Beta band 13-30Hz and Gamma band 30-50Hz, extract differential entropy frequency domain features for a single frequency band to obtain 16 differential entropy (DE) features, extract differential entropy features for five frequency bands, and extract 16*5=80 differential entropy (DE) features for each group of data.

[0070] S4. Use PCA method to reduce the dimension of features.

[0071] S41. First, the data is decentralized and normalized by subtracting the mean value from each variable. This way, the center of the data can be returned to the zero point, that is, the mean is 0.

[0072] S42. Calculate the covariance matrix of the centered sample X.

[0073] S43. Calculate the eigenvalues ​​of the covariance matrix and arrange them in descending order.

[0074] S44. Calculate the eigenvector corresponding to the eigenvalue, and take the eigenvectors corresponding to the first p eigenvalues ​​to form a projection matrix.

[0075] S45. Convert the data into the new space constructed by the above projection matrix, so as to achieve the purpose of dimensionality reduction.

[0076] S5. The data after dimensionality reduction were divided into training set and test set according to ten-fold cross validation, and the data were classified by SVM to obtain the average classification accuracy of the six subjects in Delta, Theta, Alpha, Beta, Gamma and five frequency bands.

[0077] Embodiment 2:

[0078] See also Figure 7-10 The EEG-based music-induced emotion recognition system includes a data acquisition unit 1, the output signal of the data acquisition unit 1 is connected to a system control unit 2, and the output signal of the system control unit 2 is connected to a data processing unit 3.

[0079] Furthermore, the data acquisition unit 1 includes an acquisition module 11 , the output end signal of the acquisition module 11 is connected to the receiving module 12 , the output end signal of the receiving module 12 is connected to the uploading module 13 , and the output end signal of the uploading module 13 is connected to the storage module 14 .

[0080] Furthermore, the system control unit 2 includes a touch module 21 , the output end signal of the touch module 21 is connected to the sensor module 22 , the output end signal of the sensor module 22 is connected to the feedback module 23 , and the output end signal of the feedback module 23 is connected to the output module 24 .

[0081] Furthermore, the data processing unit 3 includes a calling module 31 , the output end signal of the calling module 31 is connected to the analyzing module 32 , the output end signal of the analyzing module 32 is connected to the processing module 33 , and the output end signal of the processing module 33 is connected to the display module 34 .

[0082] The EEG-based music-induced emotion recognition method and system, although the classification results using power spectral density as a feature are different from those using differential entropy as a feature, the classification effect of differential entropy features is better in the Belta band, Gamma band and all bands. Most emotional features exist in high-frequency EEG signals, so the recognition accuracy using Gamma band features is the highest, followed by Beta band, while Alpha, Theta and Delta band features cannot correctly identify emotions. Emotion recognition based on EEG signals is a method that is very sensitive to personalization. For different subjects, their different physiological and psychological conditions, different perceptions of stimulus materials, etc., will cause large differences in effects.

[0083] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A music-induced emotion recognition method based on EEG, characterized in that: The following steps are involved: S1, design the experimental paradigm and collect EEG data; S11. Design a BCI experimental paradigm based on music stimulation emotions. Music stimulation includes two types: joyful and sad. S12. Use E-prime to design the experimental task interface, including the gaze interface, music stimulation interface, feedback interface, and rest interface; S13. Before the experiment begins, instructions will be displayed in the center of the screen to let the subjects fully understand the experimental rules. Before the formal data collection, the subjects should adjust their emotions and be in a state of physical and mental relaxation, wear headphones and be ready for the experiment; S14. At the beginning of the experiment, a plus sign "+" will appear in the center of the screen for 3 seconds, prompting the subjects to prepare themselves for the mood. Then, when the music symbol appears, the music will start playing. The music will play for 60 seconds, and then the screen will prompt "Please rate the type of music" for 5 seconds. If the subject thinks the music is positive, press the "9" key, if the subject thinks the music is negative, press the "1" key, if he cannot judge, press the "5" key, and finally, an asterisk "*" will be displayed in the center of the screen for 2 seconds. During this period, the subject will rest and adjust his mood. S2, preprocessing the collected EEG signals; S21. Three sets of EEG data of the subjects were obtained by measuring electrodes. The cheerful music and sad music were sorted into two data sets based on the moment label data. The original EEG signal was first processed using a 50 Hz notch filter. S22, using a 6th order Butterworth filter to filter the EEG signal in the range of 1-50 Hz, segmenting the EEG data according to the time label of the measuring electrode, taking the 60-second time period of music playing for processing, and then performing time-frequency analysis on all electrodes, drawing the power spectrum density diagram of the 16 electrodes of the two types of data and the power spectrum density difference diagram of the two types of music data; S23. Based on the power spectrum density diagram and the power spectrum difference diagram of the two types of emotions, the difference changes of the two types of emotions in the 15-25Hz frequency band of electrode P4 were obtained. The 9 positive emotion experimental data and the 9 negative emotion experimental data of electrode P4 at the superposition frequency of 15-25Hz were taken as two samples. A paired t-test was performed on the two samples, and P<0.05 was obtained, proving the difference between the two types of emotions. S24, consider adjusting the bandpass filter range to 15-25Hz, after determining the frequency range of energy change, set the data energy in the 60-second music playback after filtering to A, set the energy of the reference time period before the prompt to R, and the energy fluctuation percentage D of the 60-second data relative to the energy before the prompt occurs to S3, extract features from 60 seconds of music playing; S31, converting the time domain information into time-frequency information, performing short-time Fourier transform on each group of each pilot signal, using Hanning window as the window function, and setting the window size to 1 second, the time domain sampling frequency to 250Hz, the frequency domain sampling frequency to 512Hz, and the emotional information contained in the characteristics of each frequency band to be different. In this way, the time domain information within 1 second will be converted to the frequency domain by a Fourier transform, so that for each type of emotional data, 9*60*250=135000 sampling points are obtained; S32, take out the five frequency bands closely related to physiological and psychological activities, namely Delta frequency band 1-4Hz, Theta frequency band 4-8Hz, Alpha frequency band 8-13Hz, Beta frequency band 13-30Hz and Gamma frequency band 30-50Hz, extract differential entropy frequency domain features for a single frequency band to obtain 16 differential entropy (DE) features, extract differential entropy features for five frequency bands, and extract 16*5=80 differential entropy (DE) features for each group of data; S4. Use PCA method to reduce the dimension of features; S41. First, the data is decentralized and normalized by subtracting the mean value from each variable, so that the center of the data can be returned to the zero point, that is, the mean is 0; S42, calculating the covariance matrix of the centered sample X; S43, calculating the eigenvalues ​​of the covariance matrix and arranging them in descending order; S44, calculating the eigenvector corresponding to the eigenvalue, taking the eigenvectors corresponding to the first p eigenvalues ​​to form a projection matrix; S45, converting the data into a new space constructed by the projection matrix, so as to achieve the purpose of dimensionality reduction; S5. The data after dimensionality reduction were divided into training set and test set according to ten-fold cross validation, and the data were classified by SVM to obtain the average classification accuracy of the six subjects in Delta, Theta, Alpha, Beta, Gamma and five frequency bands.

2. The EEG-based music-induced emotion recognition method according to claim 1, characterized in that: In step S1, the EEG signal acquisition device includes 16 electrodes: F3, FZ, F4, FC1, FC2, FC5, FC6, C3, CZ, C4, CP1, CP2, CP5, CP6, P3, PZ, P4, with a sampling rate of 250 Hz. According to the international standard lead 10 / 20 system setting, the electrodes to be measured are placed on the scalp of the subject, the reference electrode and the ground electrode are located at the earlobe and forehead, and the electrode output end is connected to the wireless EEG acquisition system.

3. The EEG-based music-induced emotion recognition method according to claim 1, characterized in that: In step S1, each type of emotional music appears randomly three times in a set of experiments, and each subject completes three sets of experiments in total, with a twenty-minute rest between each set of experiments.

4. The EEG-based music-induced emotion recognition method according to claim 1, characterized in that: In step S1, six one-minute music segments are used as stimulus materials, wherein three segments reflect joyful and happy emotions, and three segments reflect sad and painful emotions. The music is played randomly, and there is a 5-second feedback time after each segment of music is played for 60 seconds. If the subject thinks that the music is positive and cheerful music, he presses the "9" key; if he thinks that the music is negative and sad music, he presses the "1" key; if he cannot judge the type of music, he presses the 5 key. Each subject is required to do three sets of experiments, and each set of experiments is separated by twenty minutes.

5. An EEG-based music-induced emotion recognition system applied to the EEG-based music-induced emotion recognition method according to any one of claims 1 to 4, comprising a data acquisition unit 1, characterized in that: The output end signal of the data acquisition unit 1 is connected to the system control unit 2 , and the output end signal of the system control unit 2 is connected to the data processing unit 3 .

6. The EEG-based music-induced emotion recognition system according to claim 5, characterized in that: The data acquisition unit 1 comprises an acquisition module 11 , the output end signal of the acquisition module 11 is connected to a receiving module 12 , the output end signal of the receiving module 12 is connected to an uploading module 13 , and the output end signal of the uploading module 13 is connected to a storage module 14 .

7. The EEG-based music-induced emotion recognition system according to claim 5, characterized in that: The system control unit 2 includes a touch module 21 , the output end signal of the touch module 21 is connected to the sensor module 22 , the output end signal of the sensor module 22 is connected to the feedback module 23 , and the output end signal of the feedback module 23 is connected to the output module 24 .

8. The EEG-based music-induced emotion recognition system according to claim 5, characterized in that: The data processing unit 3 comprises a calling module 31 , the output end signal of the calling module 31 is connected to the analyzing module 32 , the output end signal of the analyzing module 32 is connected to the processing module 33 , and the output end signal of the processing module 33 is connected to the display module 34 .

Images