AegisVox: Factorized Privacy and Clinical Preservation for Long-Form Speech

Key points

• Clinical speech reveals identity through both spoken identifiers and voice, yet its pauses, delivery, and speaker roles can also carry useful clinical evidence.1
• Protecting that speech therefore requires more than preserving its words.1
• 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.1
• Text-to-speech rendering realizes 535/601 planned replacements versus 453/601 for the codec-based speech editor.1

Sources (1)

• [1]AegisVox: Factorized Privacy and Clinical Preservation for Long-Form Speech
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 10, 06:53 AM

  We present AegisVox, a factorized framework that assigns consistent surrogate voices and edits identifier-bearing turns using document-consistent replacements.

  Clinical speech reveals identity through both spoken identifiers and voice, yet its pauses, delivery, and speaker roles can also carry useful clinical evidence.

Extractive summary: sentences quoted from the sources.