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

Noise suppression models running in production environments are commonlytrained on publicly available datasets. However, this approach leads toregressions due to the lack of training/testing on representative customerdata. Moreover, due to privacy reasons, developers cannot listen to customercontent. This `ears-off' situation motivates augmenting existing datasets in aprivacy-preserving manner. In this paper, we present Aura, a solution to makeexisting noise suppression test sets more challenging and diverse while beingsample efficient. Aura is `ears-off' because it relies on a feature extractorand a metric of speech quality, DNSMOS P.835, both pre-trained on data obtainedfrom public sources. As an application of Aura, we augment the INTERSPEECH 2021DNS challenge by sampling audio files from a new batch of data of 20K cleanspeech clips from Librivox mixed with noise clips obtained from AudioSet. Auramakes the existing benchmark test set harder by 0.27 in DNSMOS P.835 OVLR (7%),0.64 harder in DNSMOS P.835 SIG (16%), increases diversity by 31%, and achievesa 26% improvement in Spearman's rank correlation coefficient (SRCC) compared torandom sampling. Finally, we open-source Aura to stimulate research of test setdevelopment.

Aura: Privacy-preserving Augmentation to Improve Test Set Diversity in Speech Enhancement