A Tutorial on Linear Predictive Analysis of Speech Signal

ITAKURA Fumitada

Bibliographic Information

Other Title

音声の線形予測分析

Description

In this tutorial, the fundamental concept of the linear predictive analysis of speech is introduced and its development is briefly reviewed in accordance with the authors's experiences during the past 30 years. At first, the ML(maximum likelihood estimation) for the all-pole spectrum model of speech signal is described and the spectral matching measure used in the ML method is discussed. Next, the partial autocorrelation analysis(PARCOR)is explained based on the minimum mean square method. The stability conditions for the all-pole filter is expained and is proved mathematically. As an application of the prediction analysis, the modified autocorrelation method for pitch extraction is briefly described. Lastly, the line spectrum representation of the LPC coefficients is discussed and the algorithm for extraction of the LSP parameters is described.

Journal

Technical report of IEICE. DSP

Technical report of IEICE. DSP 98 (261), 25-32, 1998-09-10

The Institute of Electronics, Information and Communication Engineers

Keywords

Details 詳細情報について

CRID: 1571417127416530816

NII Article ID: 110003279914

NII Book ID: AN10060786

ISSN: 09135685

Text Lang: ja

Data Source

CiNii Articles

Export

Report a problem