An embedded-oriented sound classification system using reservoir computing

Abstract

Although deep neural networks (DNNs) have achieved state-of-the-art results in sound classification tasks in recent years, DNNs require high computational costs, and therefore implementing DNN-based sound classification systems for embedded systems is difficult. This study aims to realize high-speed and low-power sound classification hardware and proposes an embedded-oriented sound classification system using reservoir computing.

IEICE Technical Committee on Smart Info-Media Systems