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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In recent years, deep networks have led to dramatic improvements in speechenhancement by framing it as a data-driven pattern recognition problem. In manymodern enhancement systems, large amounts of data are used to train a deepnetwork to estimate masks for complex-valued short-time Fourier transforms(STFTs) to suppress noise and preserve speech. However, current maskingapproaches often neglect two important constraints: STFT consistency andmixture consistency. Without STFT consistency, the system's output is notnecessarily the STFT of a time-domain signal, and without mixture consistency,the sum of the estimated sources does not necessarily equal the input mixture.Furthermore, the only previous approaches that apply mixture consistency usereal-valued masks; mixture consistency has been ignored for complex-valuedmasks. In this paper, we show that STFT consistency and mixture consistency canbe jointly imposed by adding simple differentiable projection layers to theenhancement network. These layers are compatible with real or complex-valuedmasks. Using both of these constraints with complex-valued masks provides a 0.7dB increase in scale-invariant signal-to-distortion ratio (SI-SDR) on a largedataset of speech corrupted by a wide variety of nonstationary noise across arange of input SNRs.

Differentiable Consistency Constraints for Improved Deep Speech Enhancement