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

Speech enhancement and separation are two fundamental tasks for robust speechprocessing. Speech enhancement suppresses background noise while speechseparation extracts target speech from interfering speakers. Despite a greatnumber of supervised learning-based enhancement and separation methods havingbeen proposed and achieving good performance, studies on applyingself-supervised learning (SSL) to enhancement and separation are limited. Inthis paper, we evaluate 13 SSL upstream methods on speech enhancement andseparation downstream tasks. Our experimental results on Voicebank-DEMAND andLibri2Mix show that some SSL representations consistently outperform baselinefeatures including the short-time Fourier transform (STFT) magnitude and logMel filterbank (FBANK). Furthermore, we analyze the factors that make existingSSL frameworks difficult to apply to speech enhancement and separation anddiscuss the representation properties desired for both tasks. Our study isincluded as the official speech enhancement and separation downstreams forSUPERB.

Investigating self-supervised learning for speech enhancement and separation