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

Existing approaches for replay and synthetic speech detection still lackgeneralizability to unseen spoofing attacks. This work proposes to leverage anovel model structure, so-called Res2Net, to improve the anti-spoofingcountermeasure's generalizability. Res2Net mainly modifies the ResNet block toenable multiple feature scales. Specifically, it splits the feature maps withinone block into multiple channel groups and designs a residual-like connectionacross different channel groups. Such connection increases the possiblereceptive fields, resulting in multiple feature scales. This multiple scalingmechanism significantly improves the countermeasure's generalizability tounseen spoofing attacks. It also decreases the model size compared toResNet-based models. Experimental results show that the Res2Net modelconsistently outperforms ResNet34 and ResNet50 by a large margin in bothphysical access (PA) and logical access (LA) of the ASVspoof 2019 corpus.Moreover, integration with the squeeze-and-excitation (SE) block can furtherenhance performance. For feature engineering, we investigate thegeneralizability of Res2Net combined with different acoustic features, andobserve that the constant-Q transform (CQT) achieves the most promisingperformance in both PA and LA scenarios. Our best single system outperformsother state-of-the-art single systems in both PA and LA of the ASVspoof 2019corpus.

Replay and Synthetic Speech Detection with Res2net Architecture