词错误率够用吗？用实体感知指标重新思考语音隐私评估

Is Word Error Rate Enough? Rethinking Privacy Evaluation in Speech with Entity-Aware Metrics

AI 导读

研究将 NLP 领域的实体感知隐私指标引入语音隐私领域，评估两种混淆技术对语音内容（尤其是命名实体）的保护效果，并指出 WER 不足以准确量化隐私水平。实验显示，在实体丰富数据上微调能提升部分实体类别的攻击性能，但对其他类别无效。作者据此给出建议：指标选择应取决于混淆方法是否保留时间对齐。

正文

View PDF HTML (experimental)

Abstract:As the use of smart devices continues to increase, their potential to capture sensitive speech content raises growing privacy concerns. It is therefore critical to develop techniques that prevent information leakage while preserving the utility of the audio, and evaluation metrics that accurately quantify the level of privacy without overestimating it. In this work, we evaluate the effectiveness of two obfuscation techniques in protecting speech content, with particular emphasis on named entities, by adapting entity-aware privacy metrics from the Natural Language Processing field to the speech privacy domain. Further, we investigate several attack scenarios and show that fine-tuning on entity-rich data improves attack performance for some entity categories but not others. Finally, we provide guidance on metric selection based on whether the obfuscation method preserves temporal alignment.

Subjects:	Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Multimedia (cs.MM); Sound (cs.SD)
Cite as:	arXiv:2610.08831 [eess.AS]
	(or arXiv:2610.08831v1 [eess.AS] for this version)
	https://doi.org/10.48550/arXiv.2610.08831 arXiv-issued DOI via DataCite

Submission history

From: Anjana Rajasekhar [view email]
[v1] Sun, 27 Sep 2026 11:50:54 UTC (92 KB)

来源：arXiv：cs.CL · arxiv.org