Télécom Paris


GET
EURASIP
HINDAWI
Echo Reduction Based on Speech Codec Parameters
Author(s)
  • Emmanuel Thepie, (BenQ Mobile)
  • Christophe Beaugeant, (Siemens)
  • Hervé Taddei, (Siemens)
  • Nicolas Dütsch, (TU München)
  • Dominique Pastor, (ENST Bretagne)
Topics
  • Adaptive filtering algorithms and structures for echo and noise control
Get the paper in PDF format
 
Acrobat Reader (version 5 minimum) is necessary to read this document.

Abstract
To improve speech quality, disturbing background noise and acoustic echo are attenuated by using signal processing techniques prior to speech encoding. A different approach consists of embedding and performing noise reduction and/or echo cancellation into the speech codec. This embedding decreases the complexity and allows integration of noise reduction and echo cancellation in the network without adding any delay or creating the so-called tandem effect. In this paper we propose a solution based on this principle. We introduce an innovative estimation of the signal to echo ratio based on a linear model of the fixed gain parameters of the speech codec. Listening test results validate the good quality for such a low complexity system.

©2006 Télécom Paris/TSI
Edition : Télécom Paris -- 2006