DPCCN: Densely-Connected Pyramid Complex Convolutional Network for Robust Speech Separation And Extraction

TL;DR

This paper introduces DPCCN, a robust time-frequency domain speech separation and extraction network that outperforms existing time-domain methods, especially in cross-domain and noisy environments, by using a densely-connected pyramid structure and a novel speaker encoder.

Contribution

The paper presents a novel densely-connected pyramid complex convolutional network (DPCCN) for robust speech separation and extraction, including a new speaker encoder and a Mixture-Remix adaptation method for cross-domain tasks.

Findings

DPCCN outperforms time-domain methods in robustness and accuracy.

Mixture-Remix fine-tuning significantly improves cross-domain speech extraction.

DPCCN achieves around 3.5 dB SISNR improvement in cross-domain tests.

Abstract

In recent years, a number of time-domain speech separation methods have been proposed. However, most of them are very sensitive to the environments and wide domain coverage tasks. In this paper, from the time-frequency domain perspective, we propose a densely-connected pyramid complex convolutional network, termed DPCCN, to improve the robustness of speech separation under complicated conditions. Furthermore, we generalize the DPCCN to target speech extraction (TSE) by integrating a new specially designed speaker encoder. Moreover, we also investigate the robustness of DPCCN to unsupervised cross-domain TSE tasks. A Mixture-Remix approach is proposed to adapt the target domain acoustic characteristics for fine-tuning the source model. We evaluate the proposed methods not only under noisy and reverberant in-domain condition, but also in clean but cross-domain conditions. Results show that for both speech separation and extraction, the DPCCN-based systems achieve significantly better performance and robustness than the currently dominating time-domain methods, especially for the cross-domain tasks. Particularly, we find that the Mixture-Remix fine-tuning with DPCCN significantly outperforms the TD-SpeakerBeam for unsupervised cross-domain TSE, with around 3.5 dB SISNR improvement on target domain test set, without any source domain performance degradation.