English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Speech enhancement employing deep neural networks (DNNs) for denoising arecalled deep noise suppression (DNS). During training, DNS methods are typicallytrained with mean squared error (MSE) type loss functions, which do notguarantee good perceptual quality. Perceptual evaluation of speech quality(PESQ) is a widely used metric for evaluating speech quality. However, theoriginal PESQ algorithm is non-differentiable, and therefore cannot directly beused as optimization criterion for gradient-based learning. In this work, wepropose an end-to-end non-intrusive PESQNet DNN to estimate the PESQ scores ofthe enhanced speech signal. Thus, by providing a reference-free perceptualloss, it serves as a mediator towards the DNS training, allowing to maximizethe PESQ score of the enhanced speech signal. We illustrate the potential ofour proposed PESQNet-mediated training on the basis of an already strongbaseline DNS. As further novelty, we propose to train the DNS and the PESQNetalternatingly to keep the PESQNet up-to-date and perform well specifically forthe DNS under training. Our proposed method is compared to the same DNS trainedwith MSE-based loss for joint denoising and dereverberation, and theInterspeech 2021 DNS Challenge baseline. Detailed analysis shows that thePESQNet mediation can further increase the DNS performance by about 0.1 PESQpoints on synthetic test data and by 0.03 DNSMOS points on real test data,compared to training with the MSE-based loss. Our proposed method alsooutperforms the Challenge baseline by 0.2 PESQ points on synthetic test dataand 0.1 DNSMOS points on real test data.

Deep Noise Suppression Maximizing Non-Differentiable PESQ Mediated by a Non-Intrusive PESQNet