Joint NN-Supported Multichannel Reduction of Acoustic Echo, Reverberation and Noise

Abstract

We consider the problem of simultaneous reduction of acoustic echo, reverberation and noise. In real scenarios, these distortion sources may occur simultaneously and reducing them implies combining the corresponding distortion-specific filters. As these filters interact with each other, they must be jointly optimized. We propose to model the target and residual signals after linear echo cancellation and dereverberation using a multichannel Gaussian modeling framework and to jointly represent their spectra by means of a neural network. We develop an iterative block-coordinate ascent algorithm to update all the filters. We evaluate our system on real recordings of acoustic echo, reverberation and noise acquired with a smart speaker in various situations. The proposed approach outperforms in terms of overall distortion a cascade of the individual approaches and a joint reduction approach which does not rely on a spectral model of the target and residual signals.

U-Net Based Direct-Path Dominance Test for Robust Direction-of-Arrival Estimation

“…Therefore, it is reasonable to establish a network structure based on U-net. As noted in the field of speech enhancement [18], it is difficult to alleviate the influence of both reverberation and interference. Thus extracting the direct-path speech using Unet in a straightforward manner is not a proper choice.…”

Deep Residual Echo Suppression with A Tunable Tradeoff Between Signal Distortion and Echo Suppression

“…2514/17). memory (LSTM) networks to jointly obtain echo cancellation and to suppress noises and reverberations [5]. Lee et al [6] cascaded a fully-connected neural network (FCNN) after a linear acoustic echo suppressor (AES) and evaluated the objective gain between the spectra amplitudes of the near-end and AES output signals.…”