国外高校老师编写的语音信号特征提取程序，非常实用初学者研究-　[E, V, A, P] = analysis(x, N, U, M) extracts vocoder parameters E, V, A,

and P from the speech signal x on a frame by frame basis. N is the

analysis frame length, U is the update length, and M is the order of the

linear prediction analysis. The return arguments are the extracted

parameters, where E, V, and P are vectors with "n_frames" (the number of

analysis frames, see code below) elements, A is a (n_frames, M+1) size

matrix (includes the leading 1 of the polynomial).