Identification of Pathological Voices based on LPCC and MFCC
GU Ji-huaa · Communications technology · 2012
The choice of effective feature parameters is the key in the automatic detection of normal and pathological voices,then these parameters are optimized to obtain easy and achievable parameters.Meanwhile,the selection of an appropriate model could achieve the best recognition rate.In order to detect normal and pathological voices instantly and conveniently,linear prediction cepstral coefficients(LPCC) and Mel cepstral coefficients(MFCC) are proposed to identify them respectively by using dynamic time warping(DTW) model.The experimental results show that the correct recognition rate could reach 90%,and MFCC method is better than LPCC.