Algorithms And Software For Predictive And Perceptual Modeling Of Speech

This book attempts to bridge the knowledge gap between the old perceptual methods for speech coding and some of the latest techniques employed for embedding perceptual metrics in vocoders. The application of perceptual models in speech coding began receiving attention during the late nineteen seventies. Methods that exploit the masking properties of the human ear in speech coding standards are largely based on relatively old concepts introduced by Schroeder and Atal in 1979. This book also includes several MATLAB examples that facilitate readers to have hands-on experience with some of the latest perceptual models for speech coding. In Chapter 2, a review of various linear prediction (LP) based speech analysis and synthesis methods is presented. In Chapter 3, we focus on some of the existing perceptual methods in both narrowband and wideband speech coding standards. Limitations associated with these models are described using MATLAB examples. Recent efforts to embed perceptual metrics in wideband speech coding are also included.