Efficient Acoustic Echo Cancellation joint with noise reduction framework

Abstract

Traditionally, acoustic echo problem was solved by employing large scale digital signal processors. However, such system is costly and it should be implemented in advance with audio devices. With the advancement of computational speed of personal computer, researchers are exploring ways to design Acoustic Echo Cancellation (AEC) software that is able to reduced echoes resulting from the acoustic coupling between loudspeaker and microphone. Although numerous algorithms have been developed in recent years, the existing AEC algorithms are unable to tackle the issues for devices that have different sampling rate. Besides the adaptive filter that used in AEC system, re-sampling algorithms that is able to match the sampling rate of the input signals to the AEC system, and synchronization controller between speaker signal and microphone signal is also required. Therefore, in this paper a new AEC system framework has been proposed that can handle the mismatch in the sampling rate of the input signals and generate a balanced sampling rate output.