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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

This paper describes the deepfake audio detection system submitted to theAudio Deep Synthesis Detection (ADD) Challenge Track 3.2 and gives an analysisof score fusion. The proposed system is a score-level fusion of several lightconvolutional neural network (LCNN) based models. Various front-ends are usedas input features, including low-frequency short-time Fourier transform andConstant Q transform. Due to the complex noise and rich synthesis algorithms,it is difficult to obtain the desired performance using the training setdirectly. Online data augmentation methods effectively improve the robustnessof fake audio detection systems. In particular, the reasons for the poorimprovement of score fusion are explored through visualization of the scoredistributions and comparison with score distribution on another dataset. Theoverfitting of the model to the training set leads to extreme values of thescores and low correlation of the score distributions, which makes score fusiondifficult. Fusion with partially fake audio detection system improves systemperformance further. The submission on track 3.2 obtained the weighted equalerror rate (WEER) of 11.04\%, which is one of the best performing systems inthe challenge.

Deepfake Detection System for the ADD Challenge Track 3.2 Based on Score Fusion