Speech Emotion Recognition under Noise Background
-
摘要: 为有效实现含噪声的语音信号情感类型的识别,提出了一种基于抗噪声的模糊最小二乘支持向量机的语音情感识别新方法.该方法先提取情感语音的韵律特征和音质特征,再利用模糊最小二乘支持向量机构建最优分类超平面,实现生气、高兴、悲伤和惊奇4种情感类型的识别.试验结果表明,与其他多种语音情感识别方法相比,在不同信噪比下,新方法的情感正确识别率较高,抗噪声识别效果较好,表明了其有效性.
-
关键词:
- 模糊最小二乘支持向量机 /
- 韵律特征 /
- 音质特征 /
- 情感识别
Abstract: To efficiently implement emotion recognition in noise speech signals,a new method of speech emotion recognition based on FLSSVM (fuzzy least squares support vector machines) was proposed. With this method,the prosody and voice quality features are extracted from emotional speech,and then FLSSVM is used to construct the optimum separating hyperplane so as to realize the recognition of four main speech emotions,i.e.,anger,happiness,sadness and surprise. Experimental results show that compared with the other methods of speech emotion recognition,FLSSVM can achieve higher accuracies and better anti-noise effects for speech emotion recognition at different levels of signal-to-noise ratio,indicating the efficiency of the proposed method. -
COWIE R,CORNELIUS R R.Describing the emotional states that are expressed in speech[J].Speech Communication,2003,40(112):5-32.[2] JUAN M M,ARANTZA A.Emotions in human and artificial intelligence[J].Computers in Human Behavior,2005,2(21):323-341.[3] BREAZEAL C.Emotion and sociable humanoid robots[J].International Journal of Human-Computer Studies,2003,1-2(59):119-155.[4] 赵力,钱向民,邹采荣,等.语音信号中的情感识别研究[J].软件学报,2001,12(07):1050-1055.ZHAO Li,QIAN Xiangmin,ZOU Cairong,et al.A study on emotional recognition in speech signal[J].Journal of Software 2001,12(07):1050-1055.[5] NEIBERG D,ELENIUS K,LASKOWSKI K.Emotion recognition in spontaneous speech using GMMs[C]//Proceedings of Ninth International Conference on Spoken Language Processing.Pittsburg:ISCA,2006:809-812.[6] 张石清,赵知劲.基于多分类器投票组合的语音情感识别[J].微电子学与计算机,2008,12:17-20.ZHAN Shiqing,ZHAO Zhijin.Speech emotion recognition based on voting combination of multiple classifiers[J].Micrcel Ectronies Computer,2008,12:17-20.[7] YOU M,CHEN C,BU J,et al.Manifolds based emotion recognition in speech[J].Computational Linguistics and Chinese Language Processing,2007,12(1):49-64.[8] TSUJINISHI D,ABE S.Fuzzy least squares support vector machines for multiclass problems[J].Neural Network,2003,16:785-792.[9] 赵力,蒋春辉,邹采荣,等.语音信号中的情感特征分析和识别的研究[J].电子学报,2004,32(4):606-609.ZHAO Li,JIANG Chunhui,ZOU Cairong,et al.A study on emotional feature analysis and recognition in speech[J].Acta Electronica Sinica,2004,32(4):606-609.[10] ANG J,DHILLON R,KRUPSKI A,et al.Prosody-based automatic detection of annoyance and frustration in human-computer dialog[C]//Proceedings of Ninth International Conference on Spoken Language Processing.Denver:ISCA,2002:2037-2039.[11] ZHANG Shiqing.Speech emotion recognition based on fuzzy least squares support vector machines[C]//Proceedings of Seventh World Congress on Intelligent Control and Automation.Chongqing:WCICA,2008:1299-1302.
点击查看大图
计量
- 文章访问数: 1654
- HTML全文浏览量: 55
- PDF下载量: 564
- 被引次数: 0