Speaker Anonymisation for Speech-based Suicide Risk Detection

2025-09-26

AI Summary

This study addresses speaker identity leakage—a critical privacy risk in adolescent suicide risk detection from speech—by proposing the first systematic voice anonymization framework tailored to this sensitive task. To preserve both privacy and diagnostic utility, we integrate traditional signal processing, neural voice conversion, and end-to-end speech synthesis into a multi-strategy anonymization pipeline, accompanied by a standardized evaluation protocol to quantify the privacy–utility trade-off. Experiments demonstrate that our method achieves near-original detection accuracy (within 2% degradation) while reducing speaker identification accuracy by over 60%, substantially outperforming single-method baselines. This work establishes the first practical solution that simultaneously ensures speaker anonymity and clinical reliability in suicide risk speech analysis, thereby enabling trustworthy voice-based mental health monitoring in high-stakes, privacy-sensitive scenarios.

Technology Category

• Natural Language Processing: Speech
• Machine Learning: Privacy
• Search and Optimization: Mixed Discrete/Continuous Search

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

Adolescent suicide is a critical global health issue, and speech provides a cost-effective modality for automatic suicide risk detection. Given the vulnerable population, protecting speaker identity is particularly important, as speech itself can reveal personally identifiable information if the data is leaked or maliciously exploited. This work presents the first systematic study of speaker anonymisation for speech-based suicide risk detection. A broad range of anonymisation methods are investigated, including techniques based on traditional signal processing, neural voice conversion, and speech synthesis. A comprehensive evaluation framework is built to assess the trade-off between protecting speaker identity and preserving information essential for suicide risk detection. Results show that combining anonymisation methods that retain complementary information yields detection performance comparable to that of original speech, while achieving protection of speaker identity for vulnerable populations.

Problem

• Speaker anonymisation for vulnerable populations' speech data
• Balancing identity protection with suicide risk detection accuracy
• Evaluating multiple anonymisation methods' impact on detection performance

Innovation

• Combining complementary anonymisation methods for protection
• Using neural voice conversion and speech synthesis techniques
• Balancing identity protection with risk detection accuracy