Dynamic Structure Linear Predictive Coder

The dynamic structure linear predictive coder (DLPC) is proposed. The analysis filter structure is dynamic, i.e., the order number is varied according to the analyzing gain. It is adjusted so as to the achieved gain is less than the reference gain value. The main advantage is that the DLPC provides low bit-rate of transmission. Also, the subjective and objective measurements show that the DLPC yields relative quality of the synthesized speech compared to that of the conventional LPC coding

I. Introduction

The goal of coding speech is to transform the raw speech signal into a more compact form, which can then be rapidly transmitted or consumed a considerably smaller memory. Generally, speech codec and speech compression are required for long distant communication, massage encryption, speech storage, and etc. For instance, in the cellular mobile system many users need to share the same frequency bandwidth. The speech compression is therefore important. Normally, the output signal obtained from a speech codec does not sound similar to the input because the speech coding is a lossy type of coding. This means that the input and the output signals can be evidently distinguished.