Yu-Xiang Wang

VoxPrivacy: A Benchmark for Evaluating Interactional Privacy of Speech Language Models

The VoxPrivacy benchmark, the large-scale training set, the fine-tuned model to foster the development of safer and more context-aware SLMs, and a viable path forward are demonstrated: by fine-tuning on a new 4,000-hour training set, improve privacy-preserving abilities while maintaining robustness.