Audio encoding for heart and breath sounds acquired with digital stethoscope

Abstract

This paper presents a real-time audio encoding system for cardiorespiratory sounds acquired with a low cost Bluetooth digital stethoscope developed by EHAS (Enlace Hispano Americano de Salud) Foundation and Fundatel Foundation. The system is currently working with a G.722 speech coder for compression and subsequent transmission of cardiorespiratory signals. However, these signals have different frequency characteristics from speech. Therefore, we sought a better adapted alternative to encode these signals optimally, with an encoder not subject to the payment of any license, with low computational cost, good quality and low bandwidth. We have evaluated the proposed solution both using objective measures such as root mean squared error and using subjective opinions from four expert physicians.