论文部分内容阅读
为了提高情感识别的正确率,针对单一语音信号特征和表面肌电信号特征存在的局限性,提出了一种集成语音信号特征和表面肌电信号特征的情感自动识别模型。首先对语音信号和表面肌电信号进行预处理,并分别提取相关的语音信号和表面肌电信号特征,然后采用支持向量机对语音信号和表面肌电信号特征进行学习,分别建立相应的情感分类器,得到相应的识别结果,最后将识别结果分别输入到支持向量机确定两种特征的权重系数,从而得到最终的情感识别结果。两个标准语情感数据库的仿真结果表明,相对于其它情感识别模型,本文模型大幅提高了情感识别的正确率,人机交互情感识别系统提供了一种新的研究工具。
In order to improve the accuracy of emotion recognition, aiming at the limitations of single speech signal features and surface EMG signal features, an automatic emotion recognition model integrating speech signal features and surface EMG features is proposed. Firstly, the speech signal and the surface electromyography signal are preprocessed, and the related speech signals and the surface electromyography signal characteristics are extracted respectively. Then, the support vector machine is used to learn the characteristics of the speech signal and the surface electromyography signal, and the corresponding affective classification The corresponding recognition results are obtained. Finally, the recognition results are respectively input to the support vector machine to determine the weight coefficients of the two features so as to obtain the final emotion recognition result. The simulation results of two standard sentiment databases show that compared with other sentiment recognition models, the proposed model greatly improves the accuracy of sentiment recognition, and human-machine interaction sentiment recognition system provides a new research tool.