基于LPMCC的语音情感识别研究

Speech recognition is the key technology of human-computer speech intelligent interaction, feature extraction is an important part of emotional speech recognition. In this paper, we use linear prediction Mel-frequency cepstral coefficients (LPMCC) to reflect speech emotion. LPMCC are spectral features, which are proposed based on Mel frequency and Linear prediction cepstral coefficients (LPCC). LPMCC combines the advantages of Mel-frequency cepstral coefficients (MFCC) and linear prediction cepstral coefficients (LPCC) well, this approach considers the human auditory characteristics and also improves the efficiency of the algorithm. In this paper, linear prediction Mel-frequency cepstral coefficients (LPMCC) were extracted from emotional sentences by using different emotional voicelibs of Chinese and English, and recognized by support vector machine (SVM). Experimental results show that the recognition rate of LPMCC is better than that of LPCC, and the recognition rate of the single Chinese or the English emotional voicelib is better than that of the mixed emotional voicelib.