A new face-to-voice deepfake technique, FOICE, can generate convincing speech from photos, challenging existing detection methods. This innovation has significant implications for voice authentication systems, exposing vulnerabilities and highlighting the need for broader, more adaptable detection strategies. #FOICE #FaceToVoiceDeepfake
Keypoints
- FOICE creates speech from images by estimating vocal traits based on facial appearance.
- The technique can fool voice-based authentication systems like WeChat Voiceprint with increasing success.
- Current deepfake detectors struggle to identify FOICE-generated voices because they lack traditional synthetic patterns.
- Retraining detectors with FOICE samples improves accuracy temporarily but reduces effectiveness against other fake types.
- The study emphasizes the need for more comprehensive datasets and proactive detection models to combat evolving deepfake methods.
Read More: https://www.helpnetsecurity.com/2025/10/30/face-to-voice-deepfakes-voice-authentication-risk/