Many practical systems are nonlinear in nature, and the Volterra series, also known as nonlinear convolution, is widely used to model these systems. For nonlinear systems with infinite memory, such a modeling approach is usually not feasible because of multiple infinite summations. In practice, the full Volterra series representation of such a system is either approximated by just a few terms, or is otherwise simplified. In an audio system, a useful approximation is to model all memoryless and dynamical nonlinear effects as a combined nonlinearity at its output. In this paper we propose a new Volterra-based structure that accommodates nonlinear systems with output nonlinearity and infinite memory. We then propose an adaptation approach to estimate the Volterra kernels based on the Least Mean Squares (LMS) approach.

Authors: Soltanmohammadi, Erfan; Painter, Christopher; Jain, Kapil
Affiliations: Marvell Semiconductor, Inc., Santa Clara, CA, USA; Marvell Semiconductor, Inc., Longmont, CO, USA(See document for exact affiliation information.)
AES Convention: 140 (May 2016) Paper Number: 9564
Publication Date: May 26, 2016 Import into BibTeX
Subject: Audio Signal Processing: Audio Applications
Permalink: https://www.aes.org/e-lib/browse.cfm?elib=18262

Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!

This paper costs $33 for non-members and is free for AES members and E-Library subscribers.

Learn more about the AES E-Library

E-Library Location: (CD 140Papers)   /conv/140/9564.pdf

Start a discussion about this paper!