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

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

Stať ve sborníku v databázi WoS či Scopus

In recent years, a number of time-domain speech separation methodshave been proposed. However, most of them are very sensitiveto the environments and wide domain coverage tasks. In thispaper, from the time-frequency domain perspective, we propose adensely-connected pyramid complex convolutional network, termedDPCCN, to improve the robustness of speech separation under complicatedconditions. Furthermore, we generalize the DPCCN to targetspeech extraction (TSE) by integrating a new specially designedspeaker encoder. Moreover, we also investigate the robustness ofDPCCN to unsupervised cross-domain TSE tasks. A Mixture-Remixapproach is proposed to adapt the target domain acoustic characteristicsfor fine-tuning the source model. We evaluate the proposedmethods not only under noisy and reverberant in-domain condition,but also in clean but cross-domain conditions. Results show that forboth speech separation and extraction, the DPCCN-based systemsachieve significantly better performance and robustness than the currentlydominating time-domain methods, especially for the crossdomaintasks. Particularly, we find that the Mixture-Remix finetuningwith DPCCN significantly outperforms the TD-SpeakerBeamfor unsupervised cross-domain TSE, with around 3.5 dB SISNR improvementon target domain test set, without any source domain performancedegradation.

In recent years, a number of time-domain speech separation methodshave been proposed. However, most of them are very sensitiveto the environments and wide domain coverage tasks. In thispaper, from the time-frequency domain perspective, we propose adensely-connected pyramid complex convolutional network, termedDPCCN, to improve the robustness of speech separation under complicatedconditions. Furthermore, we generalize the DPCCN to targetspeech extraction (TSE) by integrating a new specially designedspeaker encoder. Moreover, we also investigate the robustness ofDPCCN to unsupervised cross-domain TSE tasks. A Mixture-Remixapproach is proposed to adapt the target domain acoustic characteristicsfor fine-tuning the source model. We evaluate the proposedmethods not only under noisy and reverberant in-domain condition,but also in clean but cross-domain conditions. Results show that forboth speech separation and extraction, the DPCCN-based systemsachieve significantly better performance and robustness than the currentlydominating time-domain methods, especially for the crossdomaintasks. Particularly, we find that the Mixture-Remix finetuningwith DPCCN significantly outperforms the TD-SpeakerBeamfor unsupervised cross-domain TSE, with around 3.5 dB SISNR improvementon target domain test set, without any source domain performancedegradation.

DPCCN, Mixture-Remix, cross-domain, speech separation, unsupervised target speech extraction

DPCCN, Mixture-Remix, cross-domain, speech separation, unsupervised target speech extraction

HAN, J.; LONG, Y.; BURGET, L.; ČERNOCKÝ, J.

2023

27.05.2022

IEEE Signal Processing Society

Singapore

978-1-6654-0540-9

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

7292

7296

5

https://ieeexplore.ieee.org/document/9747340

han_icassp2022_DPCCN_Densely-Connected_Pyramid_Complex_Convolutional_Network_for_Robust_Speech_Separation_and_Extraction