Options
2018
Conference Paper
Titel
A low-complexity state-space architecture for multi-microphone acoustic echo control
Abstract
Modern hands-free communication devices, tablets, smart speakers and televisions, employ multi-microphone signal processing for speech enhancement and recognition. Given a multi-microphone (MM) setup, the signals reproduced by the loudspeaker(s) are fed back to the microphones, resulting in acoustic coupling. The most common solution to reduce acoustic echoes is to employ one echo canceler per microphone. Yet, in practice, such an implementation may not be feasible due to computational complexity constraints. Moreover, regardless of the adaptive algorithm used, audible residual echoes may remain after echo cancellation, and additional residual echo suppression is commonly required. Following up on recent work on complexity reduction of MM acoustic echo cancellation, a state-space architecture for low-complexity MM echo control is presented here, which incorporates residual echo reduction adding nearly no computational complexity.