Design And Implementation Of Dsp-based Echo Offset

Acoustic Echo Canceller (AEC) is an indispensable device applied in many communication situations, such as hands-free wireless systems, Teleconferencing etc, where a loudspeaker and a microphone are placed within an enclosure. The AEC has two main functions: eliminating the echoes due to acoustic feedback from loudspeaker to microphone and protecting the communication systems from howling and becoming unstable when this kind of feedback becomes seriously.The fundamental of AEC is cancelling the echo by employing an adaptive filter to identify and simulate the real echo paths. What makes the capability of an adaptive algorithm applied in AEC very demanding is the highly nonstationary nature of the involved signals and systems, as well as the very long filters required. The performance of an algorithm can be measured by some factors, such as the accuracy of the obtained solution with respect to the theoretically expected set up; the convergence speed; the robustness to round off error accumulation and computational complexity etc. With increasingly more signal processing power available of Digital Signal Processors (DSP) in recent years, the applications of an adaptive filter parallel to the loudspeaker -enclousure-microphone system in real time become economically feasible.In this article, a new adaptive algorithm (BEPNLMS) based on 'the sparseness of the echo path' and Block-Exact Techniques' is studied. And considering the design of the teleconferencing terminals, a kind of software and hardware implementation scheme of the BEPNLMS algorithm based on DSP (eg.TMS320C54X) is also offered. At the last, the main performance of the BEPNLMS algorithm is analyzed and discussed in order to demonstrate the feasibility and practicability of the new adaptive scheme.This paper is stated into five chapters as follows:Chapter 1 is about the general situation of the acoustic echo cancellation technology and the main content of our research.Chapter 2 introduces some adaptive technology applied in AEC, and analizes their merit and demerit.Chapter 3 proposes and studies a new adaptive Algorithm (BEPNLMS) based on the sparseness of the echo path and Block-Exact Techniques.Chapter 4 offers a kind of implementation of the AEC based on fixed-point DSP, which can be applied to a terminal of the teleconferencing.Chapter 5 analizes and evaluates the main performances of the new algorithm and gets the result as expected.