Complex-valued neural networks (CVNNs) have shown their excellent efficiency compared
to their real counter-parts in speech enhancement, image and signal processing …

Similar to humans' cognitive ability to generalize knowledge and skills, self-supervised
learning (SSL) targets discovering general representations from large-scale data. This …

Speech enhancement has benefited from the success of deep learning in terms of
intelligibility and perceptual quality. Conventional time-frequency (TF) domain methods …

The ICASSP 2023 Deep Noise Suppression (DNS) Challenge marks the fifth edition of the
DNS challenge series. DNS challenges were organized from 2019 to 2023 to foster …

In this work, we build upon our previous publication and use diffusion-based generative
models for speech enhancement. We present a detailed overview of the diffusion process …

Phase is important for perceptual quality of speech. However, it seems intractable to directly
estimate phase spectra through supervised learning due to their lack of spectrotemporal …

In this paper, we propose a transformer-based architecture, called two-stage transformer
neural network (TSTNN) for end-to-end speech denoising in the time domain. The proposed …

Adversarial loss in a conditional generative adversarial network (GAN) is not designed to
directly optimize evaluation metrics of a target task, and thus, may not always guide the …

For challenging acoustic scenarios as low signal-to-noise ratios, current speech
enhancement systems usually suffer from performance bottleneck in extracting the target …

Score-based generative models (SGMs) have recently shown impressive results for difficult
generative tasks such as the unconditional and conditional generation of natural images …