arXiv cs.AI· Praphul Singh, Akshat Agarwal, Ratul Ghosh, Aashna Kanuga, Devashish Khatwani, Anurag Dwarakanath, Sri Gadde, Krishnaram Kenthapadi·· 3 天前AI 评分23
AegisVox：面向长时语音的隐私与临床信息因子化保护框架
AegisVox: Factorized Privacy and Clinical Preservation for Long-Form Speech
AI 导读
AegisVox 是一个因子化框架，为长时临床语音分配一致的替代音色，并用文档一致的替换编辑含身份标识的话轮。在 390 段公开模拟或非患者对话（共 63.6 小时）上，TTS 渲染实现 535/601 次计划替换，基于 codec 的语音编辑器为 453/601。但替换实现更好可能同时伴随更差的临床或说话人关联测量，AegisVox 支持针对具体用途评估这些权衡，而非保证身份标识移除或临床安全。
正文 · 原文
Abstract:Clinical speech reveals identity through both spoken identifiers and voice, yet its pauses, delivery, and speaker roles can also carry useful clinical evidence. Protecting that speech therefore requires more than preserving its words. We present AegisVox, a factorized framework that assigns consistent surrogate voices and edits identifier-bearing turns using document-consistent replacements. Across 390 public simulated or non-patient conversations totaling 63.6 hours, a matched comparison of two speech renderers separates replacement realization from retained identifiers, speaker linkage, and clinical preservation. Text-to-speech rendering realizes 535/601 planned replacements versus 453/601 for the codec-based speech editor. However, similar aggregate clinical preservation conceals more model-labeled contradictions, and conversation-wide speaker scores conceal greater edited-speech linkage under a score-direction sensitivity analysis. Fresh transcription of original audio provides a reference for interpreting the clinical measurement chain; an independent judge tests sensitivity to model-based assessment. Better replacement realization can therefore coexist with worse clinical or speaker-linkage measurements. AegisVox supports use-specific evaluation of those tradeoffs, not a guarantee of identifier removal or clinical safety.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.14643 [cs.AI] |
| (or arXiv:2610.14643v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.14643 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Praphul Singh [view email]
[v1]
Sat, 10 Oct 2026 06:53:23 UTC (1,472 KB)
来源：arXiv cs.AI · arxiv.org