DocumentCode
3326003
Title
Performance improvement of acoustic ECHO cancellers in the presence of external volume change
Author
Leblanc, Wilf ; Nemer, Elias
Author_Institution
Broadcom Corp., Irvine, CA, USA
fYear
2012
fDate
12-14 Jan. 2012
Firstpage
123
Lastpage
126
Abstract
This paper presents a new method for improving the performance of an acoustic echo canceller operating in a hands-free telephony context, where the terminal connects to an external amplifier whose volume can be changed arbitrarily. This change would normally results in the canceller loosing convergence, and temporarily causes an echo to be heard by the far side. The proposed method provides an effective way to quickly compensate, by continuously tracking key measures across the frequency bands, and adjusting the coefficients as well as the output of the canceller´s filter, in order to minimize the level and duration of the residual echo. A subband architecture consisting of two sets of cancellers is considered. Changes in external volume (or tone) are translated into a complex gain (or set of gains) and applied to the output or the coefficients of the foreground canceller filter, according to the optimality criterion of minimizing the mean square error (MSE) at the canceller output. The algorithm uses various metrics to ensure the validity of the computed gains and the overall canceller´s stability. Listening tests and perceptual evaluation of speech quality (PESQ) score done in a real-life setting show the method results in a noted improvement of the overall perceptual performance of the canceller.
Keywords
acoustic filters; acoustic signal processing; audio signal processing; echo suppression; mean square error methods; telephony; acoustic echo canceller; canceller filter; external amplifier; external volume change; hands free telephony context; mean square error minimisation; Computer architecture; Context; Echo cancellers; Filter banks; Gain; Speech; AEC; External Volume Change;
fLanguage
English
Publisher
ieee
Conference_Titel
Emerging Signal Processing Applications (ESPA), 2012 IEEE International Conference on
Conference_Location
Las Vegas, NV
Print_ISBN
978-1-4673-0899-1
Type
conf
DOI
10.1109/ESPA.2012.6152461
Filename
6152461
Link To Document