Real-Time Implementation of a Stereophonic Acoustic Echo Canceler

In this memo, a real-time implementation of a stereophonic acoustic echo canceler is presented. Possible applications are desktop and video conferencing, where two channel audio will enhance signal separation, especially when several talkers speak simultaneously. The stereophonic acoustic echo canceler uses the two channel fast recursive least squares adaptive fibers applied in the two-path scheme to estimate the echo signal. A subband structure was chosen to reduce the calculation complexity, stabilize the adaptive filters and to be able to implement the echo canceler on hardware with parallel DSPs. The majority of the code has been written in C, with some parts optimized in native DSP assembler, thus the current implementation should be fairly portable.