Quality Assessment of Efficient LP Analysis Methods
D. Satyanarayana, K. Satya Prasad, T. J. C. Prasad · 2007
The performance of the autocorrelation method of LP analysis has always been limited when the speech signal is corrupted by noise. In order to overcome these limitations number of efficient LP analysis techniques have been developed such as Moving Average(MA), Moving Maximum (MM) and Average threshold Method(AT). All these methods here compared with conventional Auto correlation Method(AM). The quality of speech will be measured quantitatively by various parameters such as Overall SNR, Segmental SNR, Likelihood Ratio (LLR), Log Area ratio (LAR), Itakura Saito Distortion (ISD), Weighted Spectral Slope Measure (WSS). In this paper an attempt has been made to compare the efficient LP analysis methods based on above parameters. After care full analysis it is found that the Average Threshold (AT) Method of speech processing gives better results compare to other methods.