Research On Music Reconstruction Based On Brain Wave To Alleviate Anxiety

In the process of anxiety treatment through music perception,the existing music is not targeted for patients,resulting in unsatisfactory treatment effect.In order to alleviate anxiety and create targeted music,this paper combines computer composing with emotional feedback,and proposes a method of music reconstruction based on emotional feedback.Because the accuracy of subjective judgment of emotion is not high,this paper chooses EEG analysis to identify emotional feedback.The method of music reconstruction based on EEG to alleviate anxiety is mainly to recognize the changes of EEG signals of anxious patients under different music.According to emotional feedback,obtaining music fragments that can improve the mood of anxious patients.Then,based on the music content,extracting a variety of low-dimensional feature sets which can express the basic characteristics of music,using the feature set of music to detect the similarity of Euclidean distance.Determining the Best Sequence of Composite Music Fragments Based on Similarity and ensuring the coherence and naturalness of the new track,the music stitching is smoothed.In the process of music reconstruction,accurate emotional feedback is crucial.In order to improve the accuracy of EEG recognition,this paper proposes to establish a Long Short Term Memory Network(LSTM)based on the Minimum Redundancy Maximum Relevance(mRMR)feature selection.Firstly,the feature set of brainwave frequency band power,sliding sample entropy and Alpha asymmetry index is selected by mRMR algorithm.Then the feature matrix is input into LSTM neural network for training.During the training process,a stable model is obtained by using Adam gradient optimization algorithm.The simulation results show that the mRMR algorithm can achieve higher recognition and lower computational complexity under low-dimensional conditions.The Adam algorithm converges faster and shortens the training time;the mRMR+LSTM algorithm averaging reaches With an accuracy rate of 96.8%,it has a better classification effect than other traditional classification algorithms.Based on the framework of mRMR+LSTM emotional recognition,this paper carries out music reconstruction experiment with music content similarity as the criterion.The experimental results show that participants scored above 80 on the reconstructed music.In the follow-up music therapy process,the constructed music can improve the effect of music therapy to a certain extent.It proves that music reconstruction based on EEG signal processing can improve the effect of music therapy and the feasibility and effectiveness of music reconstruction based on EEG signal processing.