A study on the combination of emotion keywords to improve the negative emotion recognition accuracy

The speech emotion recognition is one of the important research topics on the investigation of human emotion. Most of the speech emotion recognition researches used only the speech signal as the input. The recognition rate in this case is generally no high enough for practical applications. In this paper, we proposed a method which fuses the results of the emotion keyword analysis and speech emotion recognition to improve the recognition rate for recognizing negative emotion in the speech. The results revealed that our proposed method can slightly improve the correct recognition of anger emotion but can drastically reduce the false categorization of happy emotion to to anger.