Skip to main content
QUICK REVIEW

[论文解读] Channel-wise Gated Res2Net: Towards Robust Detection of Synthetic Speech Attacks

Xu Li, Xixin Wu|arXiv (Cornell University)|Jul 19, 2021
Speech Recognition and Synthesis参考文献 29被引用 6
一句话总结

本论文提出了一种新型深度学习架构——通道门控残差2Net(CG-Res2Net),通过在残差2Net模块中引入通道门控机制,动态抑制无关通道,显著提升了在自动说话人验证(ASV)中检测合成语音欺骗攻击的鲁棒性。在ASVspoof 2019逻辑访问评估中,CG-Res2Net实现了最先进性能,其中MCG-Res2Net50将EER降低了28.8%,MLCG-Res2Net50在最难检测的未见攻击(A17)上检测准确率相比Res2Net提升了6.15%。

ABSTRACT

Existing approaches for anti-spoofing in automatic speaker verification (ASV) still lack generalizability to unseen attacks. The Res2Net approach designs a residual-like connection between feature groups within one block, which increases the possible receptive fields and improves the system's detection generalizability. However, such a residual-like connection is performed by a direct addition between feature groups without channel-wise priority. We argue that the information across channels may not contribute to spoofing cues equally, and the less relevant channels are expected to be suppressed before adding onto the next feature group, so that the system can generalize better to unseen attacks. This argument motivates the current work that presents a novel, channel-wise gated Res2Net (CG-Res2Net), which modifies Res2Net to enable a channel-wise gating mechanism in the connection between feature groups. This gating mechanism dynamically selects channel-wise features based on the input, to suppress the less relevant channels and enhance the detection generalizability. Three gating mechanisms with different structures are proposed and integrated into Res2Net. Experimental results conducted on ASVspoof 2019 logical access (LA) demonstrate that the proposed CG-Res2Net significantly outperforms Res2Net on both the overall LA evaluation set and individual difficult unseen attacks, which also outperforms other state-of-the-art single systems, depicting the effectiveness of our method.

研究动机与目标

  • 提升反欺骗系统在自动说话人验证(ASV)中对未见合成语音攻击的泛化能力。
  • 解决Res2Net的局限性,即直接使用残差连接而缺乏通道注意力,导致特征选择不理想。
  • 设计一种通道门控机制,在特征融合过程中动态抑制不相关通道,以增强模型鲁棒性。
  • 评估三种不同门控机制——单组、多组和潜在空间门控——在Res2Net框架内的有效性。
  • 在未见攻击上实现优于现有最先进单模型系统的性能,尤其在最具挑战性的A17攻击上。

提出的方法

  • 提出一种新型架构CG-Res2Net,通过在特征组之间的类残差连接中插入通道门控机制,扩展了Res2Net。
  • 引入三种门控机制:单组通道门控(SCG)、多组通道门控(MCG)和多组潜在空间通道门控(MLCG),其在计算注意力权重的方式上各不相同。
  • SCG门控通过全连接层基于当前特征组单独学习通道权重。
  • MCG门控通过比较当前与下一特征组来计算注意力,实现上下文感知的抑制。
  • MLCG门控在计算门控前,将两个特征组分别投影到独立的潜在空间,从而支持更复杂、非线性的注意力学习。
  • 模型采用交叉熵损失进行训练,并在ASVspoof 2019逻辑访问(LA)基准上使用EER和t-DCF指标进行评估。

实验结果

研究问题

  • RQ1在Res2Net中引入通道门控是否能提升对ASV中未见合成语音攻击的泛化能力?
  • RQ2不同门控机制(SCG、MCG、MLCG)在提升多种攻击类型检测鲁棒性方面表现如何?
  • RQ3CG-Res2Net是否在ASVspoof 2019 LA评估集上超越Res2Net及其他最先进单模型系统,尤其在A17等难检测攻击上?
  • RQ4通道门控在特征融合过程中在多大程度上减少了无关特征的影响?

主要发现

  • MCG-Res2Net50在ASVspoof 2019 LA评估集上达到最低EER(1.78%),相比Res2Net50相对降低28.8%。
  • 在最具挑战性的未见攻击A17上,MLCG-Res2Net50检测准确率达到87.63%,相比Res2Net50提升6.15个百分点。
  • 所有三种CG-Res2Net变体(SCG、MCG、MLCG)在整体LA评估集及各类困难未见攻击上均优于Res2Net50。
  • 所提出的CG-Res2Net模型在ASVspoof 2019 LA基准上,EER和t-DCF指标均超越所有报告的单模型SOTA系统。
  • MCG-Res2Net50的t-DCF为0.052,相比Res2Net50相对降低29.7%,表明系统可靠性更优。
  • 结果证实,通道门控通过抑制不相关通道,有效增强了特征选择能力,从而提升了对未见合成语音攻击的泛化性能。

更好的研究,从现在开始

从阅读论文到最终审阅,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。