摘要
语音信号包含了人类丰富的情感信息,本文从离散情感模型出发,选择了高兴、悲伤、愤怒和害怕4种基本情感作为研究对象,利用萤火虫算法实现了支持向量机参数自动寻优,从而提高了识别的效率。为了使测试数据更据代表性,选取了中文和德文数两种据库,400个样本进行训练和测试。实验表明,对SVM分类器进行优化在一定程度上提高了情感识别率。
Speech signal contains abundant human emotional information. Starting from the discrete emotional model, this paper chooses four basic emotions: happiness, sadness, anger and fear as the research object, and uses firefly algorithm to realize automatic parameter optimization of support vector machine, thus improving the efficiency of recognition. In order to make the test data more representative, 400 samples of Chinese and German databases were selected for training and testing. Experiments show that the optimization of SVM classifier improves the emotion recognition rate to a certain extent.
作者
胡明
崔冉
郭健鹏
吴静然
翟晓东
HU Ming;CUI Ran;GUO Jian-peng;WU Jing-ran;ZHAI Xiao-dong(Xuhai College, China University of Mining and Technology, Xuzhou Jiangsu 221008)
出处
《数字技术与应用》
2019年第6期109-110,共2页
Digital Technology & Application
基金
江苏省青蓝工程资助课题
江苏省高校自然科学研究面上项目《基于深度学习的多源行为识别模型的研究》(18KJB510049)
关键词
语音信号
萤火虫算法
改进的支持向量机
情感识别
Speech signal
firefly algorithm
improved support vector machine
emotion recognition
