Attentive activation function for improving end-to-end spoofing countermeasure systems

TL;DR

This paper introduces an attention-based activation function called AReLU to enhance end-to-end spoofing detection systems by focusing on artifact-related features, demonstrating improved performance on the ASVSpoof2019 dataset.

Contribution

The paper proposes a novel attention rectified linear unit (AReLU) activation function for spoofing detection, improving feature focus and system accuracy over traditional activation functions.

Findings

AReLU improves detection accuracy on the ASVSpoof2019 dataset.

The attention mechanism enhances relevant feature contribution.

The proposed method outperforms standard activation functions.

Abstract

The main objective of the spoofing countermeasure system is to detect the artifacts within the input speech caused by the speech synthesis or voice conversion process. In order to achieve this, we propose to adopt an attentive activation function, more specifically attention rectified linear unit (AReLU) to the end-to-end spoofing countermeasure system. Since the AReLU employs the attention mechanism to boost the contribution of relevant input features while suppressing the irrelevant ones, introducing AReLU can help the countermeasure system to focus on the features related to the artifacts. The proposed framework was experimented on the logical access (LA) task of ASVSpoof2019 dataset, and outperformed the systems using the standard non-learnable activation functions.