Evaluating Parkinson's Disease Detection in Anonymized Speech: A Performance and Acoustic Analysis

2026-03-08

AI Summary

This study addresses the critical challenge of leveraging automatic speech analysis for Parkinson’s disease (PD) detection while preserving speaker privacy. The authors systematically evaluate two voice anonymization techniques—STT-TTS and kNN-based voice conversion (kNN-VC)—on two Spanish-language datasets, assessing their impact on PD classification performance. Results demonstrate that kNN-VC incurs only a modest 3–7% drop in F1 score, substantially outperforming STT-TTS, which severely degrades prosodic information. The work reveals that kNN-VC effectively preserves macro-prosodic features such as duration and fundamental frequency contours. This constitutes the first empirical evidence that carefully designed anonymization can maintain high diagnostic validity under strong privacy guarantees. Furthermore, acoustic distortion analyses offer actionable insights for advancing privacy-preserving speech-based medical technologies.

Technology Category

• Machine Learning: Privacy
• Natural Language Processing: Speech
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Security and Privacy: Privacy-enhancing technologies
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Automatic detection of Parkinson's disease (PD) from speech is a promising non-invasive diagnostic tool, but it raises significant privacy concerns. Speaker anonymization mitigates these risks, but it may suppress the pathological information necessary for PD detection. We assess the trade-off between privacy and PD detection for two anonymizers (STT-TTS and kNN-VC) using two Spanish datasets. STT-TTS provides better privacy but severely degrades PD detection by eradicating prosodic information. kNN-VC preserves macro-prosodic features such as duration and F0 contours, achieving F1 scores only 3-7\% lower than original baselines, demonstrating that privacy-preserving PD detection is viable when using appropriate anonymization. Finally, an acoustic distortion analysis characterizes specific weaknesses in kNN-VC, offering insights for designing anonymizers that better preserve PD information.

Problem

• Parkinson's disease detection
• speaker anonymization
• privacy-preserving
• speech analysis
• acoustic features

Innovation

• Parkinson's disease detection
• speaker anonymization
• prosodic preservation
• privacy-preserving AI
• acoustic distortion analysis