Research On The Method Of Generating Multimodal Physiological Signal Music

Music is a form of artistic expression that can express a person’s thoughts,emotions,and consciousness.It not only affects people’s emotions and psychology,but also affects people’s physiology.When listening to sad music,people will feel depressed and melancholy.When listening to music with bright rhythm,they will feel happy,and when listening to lively rock music,their heart rate will become faster.The feeling that the human individual produces when listening to music is the resonance between human and music.This resonance side shows that the human individual has a musical rhythm physiologically,and this rhythm is contained in breathing and heartbeat.Therefore,people’s physiological signals are collected and converted into music,and the physiological signals are recorded in the form of music.Through the analysis of music,the physiological state information of the human body can be obtained to understand the physiological state differences of the human body.The data used in this article comes from a lying detection experiment.This experiment records three physiological signals including electrocardiogram signal,skin electrical signal and expression data signal.The signals are collected to tell the truth,lie,rest and tell the truth under the specified conditions 4 Kind of physiological state.This article explores two methods for generating music with multimode physiological signals:The first one is scale-free parameter mapping: based on the scale-free nature that physiological signals and music follow together,the research on generating music using parameter mapping method is completed,and the amplitude,period and The power is mapped to the music,pitch and intensity of the notes in the music.Statistics,analysis and power-law fitting of each element of the generated music revealed that there are significant differences between the music generated by different physiological states.Through music listening experiments,it can be found that the music generated by the physiological signal can assist in distinguishing the physiological state of lying or not,and the number of physiological music listening will affect the judgment of the physiological state of the music.The second is deep learning methods: based on seq2seq(sequence to sequence)sequence-to-sequence and attention mechanism methods,the preliminary exploration of deep learning methods to generate music from physiological signals is completed.Physiological signals are input into the built seq2 seq model,which can obtain music symbols and generate music,but the results are far from our expectations and cannot generate music with distinguishable states.Through the exploration and comparison of the multi-mode physiological signal generating music method,it is found that the scale-free parameter mapping method is a better method in the current physiological signal generating music research,and deep learning cannot learn the physiological state well in the physiological signal generating music The difference between them needs further exploration and research.The research on the generation of music by physiological signals can be used as the basis and premise for analyzing and studying the physiological state expression patterns in physiological music,and it can also provide new ideas and ways for deep learning networks in music and medical clinical applications to explore deep learning.New applications in physiological signal analysis.Physiological music “audible” transforms complex information,and introduces music parameters into deep learning networks,which may provide a new idea and way to promote the development of artificial intelligence.The results can also be used to develop physiological state monitoring and feedback equipment,such as for the treatment of insomnia,pain relief,health assessment and other fields,with great economic potential.