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

To cope with reverberation and noise in single channel acoustic scenarios,typical supervised deep neural network~(DNN)-based techniques learn a mappingfrom reverberant and noisy input features to a user-defined target. Commonlyused targets are the desired signal magnitude, a time-frequency mask such asthe Wiener gain, or the interference power spectral density andsignal-to-interference ratio that can be used to compute a time-frequency mask.In this paper, we propose to incorporate multi-task learning in such DNN-basedenhancement techniques by using speech presence probability (SPP) estimation asa secondary task assisting the target estimation in the main task. Theadvantage of multi-task learning lies in sharing domain-specific informationbetween the two tasks (i.e., target and SPP estimation) and learning moregeneralizable and robust representations. To simultaneously learn both tasks,we propose to use the adaptive weighting method of losses derived from thehomoscedastic uncertainty of tasks. Simulation results show that thedereverberation and noise reduction performance of a single-task DNN trained todirectly estimate the Wiener gain is higher than the performance of single-taskDNNs trained to estimate the desired signal magnitude, the interference powerspectral density, or the signal-to-interference ratio. Incorporating theproposed multi-task learning scheme to jointly estimate the Wiener gain and theSPP increases the dereverberation and noise reduction further.

Multi-task single channel speech enhancement using speech presence probability as a secondary task training target