PulseAugur
实时 12:27:06
English(EN) Room for Error: Large-Scale Simulation of Over-the-Air Acoustic Attacks

新框架模拟语音AI声学攻击,错误率最高增加94.5% · 跟踪2个来源

研究人员开发了一个新颖的框架,用于模拟针对语音控制AI系统的空中声学攻击。该框架进行了超过800万次对抗性评估,表明声学感知可以将Whisper和wav2vec等模型的词错误率(Word Error Rate)提高高达94.5%。该研究引入了双形式信噪比(Dual-Form Signal to Noise Ratio),以更好地理解攻击的有效性和隐蔽性,旨在促进该领域更鲁棒和可验证的研究。 AI

影响 凸显了语音AI系统潜在的漏洞,需要改进安全措施以防范声学操纵。

排序理由 该集群包含一篇详细介绍新模拟框架和实验结果的学术论文。

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

新框架模拟语音AI声学攻击,错误率最高增加94.5% · 跟踪2个来源

报道来源 [2]

  1. arXiv cs.AI TIER_1 English(EN) · Andrew C. Cullen, Neil Marchant, Jiani Xie, Paul Montague, Benjamin I. P. Rubinstein ·

    Room for Error: Large-Scale Simulation of Over-the-Air Acoustic Attacks

    arXiv:2606.27701v1 Announce Type: cross Abstract: While voice control is rapidly becoming a ubiquitous vector of human-AI communication, the risks facing these systems remain poorly understood. This is, in part, a product of the difficulties in scaling strictly digital adversaria…

  2. arXiv cs.LG TIER_1 English(EN) · Benjamin I. P. Rubinstein ·

    容错性:大规模无线声学攻击模拟

    While voice control is rapidly becoming a ubiquitous vector of human-AI communication, the risks facing these systems remain poorly understood. This is, in part, a product of the difficulties in scaling strictly digital adversarial workflows to the physical world. These scale bar…