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

Distant speech recognition is being revolutionized by deep learning, that hascontributed to significantly outperform previous HMM-GMM systems. A key aspectbehind the rapid rise and success of DNNs is their ability to better managelarge time contexts. With this regard, asymmetric context windows that embedmore past than future frames have been recently used with feed-forward neuralnetworks. This context configuration turns out to be useful not only to addresslow-latency speech recognition, but also to boost the recognition performanceunder reverberant conditions. This paper investigates on the mechanismsoccurring inside DNNs, which lead to an effective application of asymmetriccontexts.In particular, we propose a novel method for automatic context windowcomposition based on a gradient analysis. The experiments, performed withdifferent acoustic environments, features, DNN architectures, microphonesettings, and recognition tasks show that our simple and efficient strategyleads to a less redundant frame configuration, which makes DNN training moreeffective in reverberant scenarios.

Automatic context window composition for distant speech recognition