Speaker Recognition System Based on weighted feature parameter
Li Zhu, Qing Sheng Yang · Physics Procedia · 2012
Speaker recognition is a kind of biometrics technology, which is very popular and widely applied. To improve the effectiveness of feature parameter, a weighted feature extraction method is proposed in this paper. It analyses each component of LPCC and finds out an average contribution sequence. Based on the sequence, LPCC is weighed by each dimension to generate a weighted feature parameter. Finally, based on VQ, a speaker recognition system which is characterized by weighted LPCC is carried out under MATLAB Environment. By experiments, speaker recognition system shows the better performance than traditional model.