Dilated FCN: Listening Longer to Hear Better
This work addresses speech enhancement for noisy audio processing, presenting an incremental improvement by optimizing network design for better performance without parameter increase.
The paper tackled the conflict between capturing long context and maintaining compact networks in speech enhancement by applying dilation operations to fully convolutional networks, resulting in models that significantly improved over baseline FCN and outperformed state-of-the-art solutions on Noisy VCTK and AzBio datasets.
Deep neural network solutions have emerged as a new and powerful paradigm for speech enhancement (SE). The capabilities to capture long context and extract multi-scale patterns are crucial to design effective SE networks. Such capabilities, however, are often in conflict with the goal of maintaining compact networks to ensure good system generalization. In this paper, we explore dilation operations and apply them to fully convolutional networks (FCNs) to address this issue. Dilations equip the networks with greatly expanded receptive fields, without increasing the number of parameters. Different strategies to fuse multi-scale dilations, as well as to install the dilation modules are explored in this work. Using Noisy VCTK and AzBio sentences datasets, we demonstrate that the proposed dilation models significantly improve over the baseline FCN and outperform the state-of-the-art SE solutions.