Least squares method for accurate speech analysis

摘要

This paper proposes a strategy to improve the performance of the linear prediction method used for speech analysis. The least squares (LS) method, one of system identification methods, is utilized to estimate the coefficients of the all-pole filter and the gain parameter. For the input estimation, a simple and effective method is proposed in which the operation of thresholding is used. It is revealed that the LS method provides a good performance independently on the pitch period of speech.