A METHOD FOR CONTINUING ADJUSTMENT OF ADAPTIVE FILTER COEFFICIENTS INVOICELESS NOISE TERMS AVAILABLE FOR ACOUSTIC ECHO CANCELER SYSTEMS

Authors
Citation
K. Fujii et J. Ohga, A METHOD FOR CONTINUING ADJUSTMENT OF ADAPTIVE FILTER COEFFICIENTS INVOICELESS NOISE TERMS AVAILABLE FOR ACOUSTIC ECHO CANCELER SYSTEMS, Electronics and communications in Japan. Part 3, Fundamental electronic science, 79(10), 1996, pp. 79-87
Citations number
10
Categorie Soggetti
Engineering, Eletrical & Electronic
ISSN journal
10420967
Volume
79
Issue
10
Year of publication
1996
Pages
79 - 87
Database
ISI
SICI code
1042-0967(1996)79:10<79:AMFCAO>2.0.ZU;2-Q
Abstract
In the acoustic echo canceller, the speaker input signal is used as a reference signal to update adaptive filter coefficients, and is someti mes composed of only the far-end background noise. This paper proposes a configuration in which the estimation error is kept constant also f or a voiceless noise period, while continuously updating the coefficie nts. The summational normalization LMS is used as the adaptive algorit hm, where the products of the residual echo and the transmitted signal (i.e., the numerators of the coefficient update resulting from the no rmalized least mean square (NLMS) algorithm), as well as the norm of t he transmitted signal (i.e., the denominator of the fore-mentioned), a re accumulated, respectively, for a certain number, and the ratio of t he two sums is used as the coefficient update. The summation number is controlled by the power of the transmitted signal. When, for example, the configuration is such that the coefficients are updated only when the sum of the norms reaches a certain value (i.e., the threshold:, t he echo cancellation is maintained at the desired value, independently of the power of the transmitted signal. The cancellation is enhanced in the period where the power of the transmitted signal is increased. In addition to the control method, which can calculate the power of th e environmental noise needed in determining the threshold, this paper analyzes the structure of the filter.