REGRESSION-BASED RESIDUAL ACOUSTIC ECHO SUPPRESSION

In this page you can find audio examples to support our paper. If you have trouble accessing the files, please send mail to Arun C. Surendran and I will promptly fix the problem.

Large room, speech in the near-end and a different speech in the far-end:

Microphone output

AEC output

RES output (output of the proposed method)

Comparison between magnitude and power regression (files correspond to Figure 5 in the paper):

AEC output

Output of RES with power regression

Output of RES with magnitude regression

RES on stereo signals:

Output after stereo AEC

Output of RES using both stereo channels