Low delay filtering for joint noise reduction and residual echo suppression

In mobile terminals speech quality is often degraded by acoustic echo and noise. Existing approaches to tackle these problems involves separated solutions. However, combined solution have been proposed recently. Low signal delay and reduced computational load are the main arguments in favor of joint noise and echo processing. Most such algorithms operate in the frequency domain. In recent works, the performance of low delay filtering structures, which could be used as alternatives to spect ral weighting, have been studied for noise reduction and echo suppression separately. In this paper, we investigate these low delay approaches for joint noise reduction and echo suppression. Results show that the approach based on the inverse discrete Fourier transform performs as well as alternative approaches but with lower signal delay and reduced computational complexity. Index Terms — Echo suppression, noise reduction, sub-band filtering, FIR filter.