Emotion Recognition Based On Hilbert-Huang Transform And Artificial Neural Network

With the continuous development of Artificial Intelligence,more and more researchers have focused on the emotional computing because of important practical application value of the effective recognition of the emotion,the emotion recognition of optical signal and speech signals has been the most important thing that researchers have concerned.Effective identification of emotional information contained in optical and acoustic signals has become the most challenging research work at present.Although in recent years,physics,psychology and neuroscience have made great progress and led the development of emotional recognition technology,the current level of technological development still has very large room to improve because of the complexity of emotion and the lag of the development of the new emotional theories of optical and speech signals.We take the development of the current emotion recognition technology by optical signal and speech signal as research background.Hilbert-Huang transform and self-organizing mapping neural network algorithm are introduced to classify the signals in the traditional emotion recognition method,and obtain effective emotion feature regions and optimizing the emotion recognition algorithm.The main research content is as follows:We study the relevant background and development history of emotion recognition,summarizes the development status of relevant technology research at home and abroad,expounds the importance and real-time nature of emotion recognition research,and analyses the application and research of Hilbert Huang transform in various scenes,expounds the significance of combining Hilbert Huang transform and self-organizing mapping neural network to research,and briefly explains the research ideas and main research contents of this paper.Emotion recognition based on the combination of optical signals and acoustic signals is studied and explored.And we study the corpus and the extraction of related characteristic parameters.Through a detailed discussion of the corpus,the value and significance of the corpus selected in this paper are illustrated.Based on the characteristics of the corpus selected in this paper,we can research the emotional analysis of non-semantic Chinese.The theoretical basis of Hilbert Huang transform and its application in optical and acoustic signal processing are studied.Starting from Hilbert-Huang transform,empirical mode decomposition is used as the method to study its decomposition method for signals,and the data get from empirical mode decomposition are analyzed.Based on the physical meaning of the selected emotion feature parameters in emotion recognition,under the condition of extracting and processing the data results of each IMF,the feature parameters obtained by empirical mode decomposition are extracted and studied.The algorithm basis and data analysis method are provided for the best suitable pattern in the emotion recognition.Emotion recognition algorithms based on Hilbert-Huang transform and self-organizing feature maps are under investigation,and the emotion recognition is carried out on a plurality of groups of signals with different characteristics.Through processing the data in the corpus,the emotion recognition accuracy results of the original data and each IMF data are obtained.Through analyzing the experimental results,it can be found that the results of a specific IMF can obtain the best emotion recognition results,and the recognition rate result of the best IMF is 12.83%higher than the recognition rate results of the original data by the result of 72.52%.And related experiments are carried out on the characteristic parameters,and the characteristic parameters are screened to obtain the best combination of the characteristic parameters and the influence of each characteristic parameter on emotion recognition.Hilbert-Huang transform can be used to process the characteristics of non-linear and non-stationary speech signals.Based on this characteristic and its self-adaptability,the emotion recognition method based on Hilbert Huang transform and self-organizing mapping neural network algorithm is studied.Based on the IMF data obtained through empirical mode decomposition,higher recognition rate results can be obtained,which proves the significance of applying Hilbert-Huang transform to emotion recognition and provides a certain reference for future research on emotion recognition based on combination of optical and acoustic signals.