Speech emotion recognition based on data mining technology

The study on the speech emotion recognition has very important realistic values in such aspects as enhancing the intelligence and humanity of computer, developing new human-machine environment and improving speech recognition results. The first goal is to search the most useful features with analyzing the features related emotions. The second gold is to find a recognition model to make use of these features. The basic course of speech emotion recognition is introduced, which includes speech signal preprocess and speech feature extraction and speech emotion recognition. After choosing the useful features such as Mel-Frequency Cepstral Coefficients (MFCC) and its transient parameters, a better performance with the application of BP neural network is obtained. Furthermore, the decision tree with multi-features is used to recognize speech emotion for comparison.