语音情绪识别(国外英文资料).docVIP

语音情感识别(国外英文资料) In October 2005, the first international conference on emotional computing and intelligent interaction was held in Beijing. Three. The research methods adopted Speech emotion recognition attention emotional information, voice of the hidden layer is a involves psychology, physiology, cross subject in the field of signal processing and pattern recognition and so on, main task is through the perception and analysis of the speech signal, stripping out the emotional expression of related acoustic features, and then identify the words of ones emotional state. In the whole identification system, the processing of emotional feature data can be critical. Typically, the number of speech emotive features is hundreds of dimensions, and the number of feature data becomes significant as the number of words increases. And we are constrained by the three-dimensional physical space that we are in, and it is very difficult to understand the data in higher dimensional space. So, faced with the huge number of high-dimensional data, how to find the character of the same emotion category data between commonness and differences between different emotional categories of feature data into a complex project. At present the characteristics of the commonly used treatment methods are actually in the traditional pattern recognition methods in use (such as support vector machines, neural network and hidden markov model, etc.), but the resulting speaker-independent speech emotion recognition performance is not ideal. Here are a few ways to illustrate the research methods of speech emotion recognition. The classification of emotions First need to study the emotions of the voice signal, according to certain standard of speech emotion features do an effective and reasonable classification, and then on the basis of different categories of study the properties of the characteristic parameters. Human emotion is quite complex, common is happiness, anger, sorrow and joy, etc. At present in