期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2012
卷号：1
期号：5
页码：65-69
出版社：Shri Pannalal Research Institute of Technolgy
摘要：This paper is mainly concerned with speech based emotion recognition. The main work is concerned with Gaussian mixture model (GMM model) which allows training the desired data set from the databases. GMM are known to capture distribution of data point from the input feature space, therefore GMM are suitable for developing emotion recognition model when large number of feature vector is available. Given a set of inputs, GMM refines the weights of each distribution through expectation-maximization algorithm. Once a model is generated, conditional probabilities can be computed for test patterns (unknown data points). Expectation maximization (EM) algorithm is used for finding maximum likelihood estimates of parameters in probabilistic models. Moreover Linear Predictive (LP) analysis method has been chosen for extracting the emotional features because it is one of the most powerful speech analysis techniques for estimating the basic speech analysis techniques for estimating the basic speech parameter such as pitch, formants, spectra, vocal tract functions and for representing speech by low bit rate transmission for storage.Speakers are made to involve in emotional conversation with the anchor, where different contextual situations are created by the anchor through the conversation to elicit different emotions from the subject, without his/her knowledge.
关键词：Speech; Gaussian Mixture Model; Vocal; Emotion Recognition; Linear predictive
